openstack/swift
Code Issues Proposed changes
ece8c28ce1
swift/CHANGELOG
Tim Burke 5f95b6b26d AUTHORS/CHANGELOG for 2.33.0 
Change-Id: I6b0a074fca4d6bd3b8b95888ae8721a62f3a6df0
2024-03-12 11:59:59 -07:00

4756 lines
186 KiB
Plaintext
Raw Blame History

swift (2.33.0, OpenStack Caracal)
* S3 API fixes:
* When the `+segments` container's storage policy differs from that of
the primary container, completed manifests are now written with the
correct policy in the primary container.
* Add basic read support for object locking. This improves
compatibility with an Ansible S3 module. Write support is not
yet implemented, so get-object-lock-configuration will always 404.
* If there's a conflict deleting the in-progress-upload marker when
completing a multipart-upload, a 503 is now returned to the client,
prompting it to retry.
* Added `Accept-Ranges: bytes` to object responses. Range requests
have always been supported; now, that support is properly advertised.
* Static large object fixes:
* Fixed a server error when handling conditional GET requests.
* Return an error if the SLO manifest could not be parsed. Previously,
a zero-byte response was returned.
* Proxy server fixes:
* Added a new `swift.proxy_logging_status` request environment key that
middlewares may use to override the logged status for a request.
* Transaction IDs are included in more error responses.
* Added a counter metric when caching shard ranges.
* The `recoverable_node_timeout` option no longer applies to
`X-Newest` GET requests.
* Improved error-handling in multi-part range responses.
* Sharding fixes:
* Prevent resets of a shard range's epoch.
* Cleaned up `X-Backend-*` headers in listing responses.
* Reduced the frequency of `Reclaimable db stuck waiting for shrinking`
messages when a root DB has been deleted but its shards have not been
shrunk away.
* The more-efficient shard range structure from the last release is now
used when fetching ranges from the backend.
* Include more information in shard-replication warnings.
* Object server fixes:
* Object POSTs and chunked PUTs are no longer accepted when the target
drive is already past its `fallocate_reserve`. DELETEs are still
allowed.
* Added the ability to configure cooperative yielding when servicing
GET responses, via the `cooperative_period` option. See the example
config for more information.
* Invalid `hashes.invalid` entries are now ignored, rather than
causing a complete partition rehash.
* Prefix-based tempurls may now be used to explore staticweb
listings within that prefix. Note that this opens a new ability
to list containers from tempurls, but only if staticweb's
`X-Container-Meta-Web-Listings` is enabled.
* When generating index pages from listings, staticweb now sends an HTML5
doctype. This makes them https://peps.python.org/pep-0503/ compliant,
allowing their continued use for simple Python package repositories.
* Per-service `auto_create_account_prefix` configuration options have
been removed. These options were deprecated in favor of `swift.conf`
configuration in Swift 2.24.0, part of the OpenStack Ussuri release.
* Added a `swift-reload` command to assist with safely reloading WSGI
servers.
* Daemons now send `STOPPING` and `RELOADING` systemd notifications
when the service is configured with `Type=notify`.
* Added more metrics to the container-server, allowing GET and PUT timings
to be broken out for listings, shard range operations, and container
creation.
* Daemons send object updates via the replication network in more cases.
* `swift-account-info` and `swift-container-info` now accept a `--sync`
flag to show information from the incoming/outgoing sync tables.
* Several fixes to prepare for Python 3.12 support. While not yet tested
in the gate, initial manual testing looks promising.
* Added support for recent versions of eventlet.
* The dark-data object audit watcher now works with sharded containers.
Previously, it would think that all data files were absent from
listings.
* Various other minor bug fixes and improvements.
swift (2.32.0, OpenStack Bobcat)
* Python 3 fixes:
* Python 3 object servers can now read unencrypted non-ASCII metadata
that was written under Python 2.
* Ssync no longer corrupts unencrypted non-ASCII metadata during
transfers.
* Fixed an encoding issue when writing non-ASCII object names to sharded
containers and shard range caching is not enabled.
* Fixed an encoding issue when handling non-ASCII account names.
* Fixed a `generator already executing` error on client disconnect.
* Suppressed `RemoteDisconnected` tracebacks.
* Metrics improvements:
* Metrics are now emitted for a variety of S3 error responses, in the
form `s3api.<status_int>.<error_class>[.<reason>]`
* Fixed an issue that prevented proxy servers from emitting metrics and
logs for backend requests made when getting account or container info.
* Account and container info metrics now include the response status code
when backend requests are made.
* Added timing metrics to the container sharder for various operations.
* Python 3.11 is now supported.
* Added the ability for reseller admins to set per-policy account quotas by
posting metadata of the form `X-Account-Quota-Bytes-Policy-<policy name>`.
* Added a `keepalive_timeout` option to the proxy server to limit how long
to wait for a client to initiate a request, separate from the general
`client_timeout` option. Note that this requires eventlet 0.34.0
(currently unreleased) or later.
* Added a `keep_cache_slo_manifest` option to the object server to better
control whether SLO manifests are dropped from the page cache.
* WSGI servers now accept a `--test-config` option that may be used to
validate configuration changes before reloading/restarting the server.
* The structure of cached shard ranges has changed, improving performance
when listing or writing to sharded containers. Note that immediately
after upgrade, the new structures will all be cache misses, which may
lead to a thundering herd problem. To avoid this, upgrade just a few
nodes first, let them service some fraction of traffic to populate the
cache, then upgrade the rest of the cluster.
* A variety of performance improvements have been made for sharded
container databases.
* Various logging and metrics improvements when talking to memcache.
* Fixed ssync's handling of timestamp offsets. Previously, this could cause
ssync to fail with a 409 Conflict, causing the transfer to fail and
preventing handoffs from clearing.
* Fixed an issue where an erasure-coded PUT could prevent other requests
from being processed when network calls rarely or never blocked.
* Fixed an issue when downloading an SLO manifest would hit a recoverable
error and attempt to resume from another node. This would manifest as
either a pyeclib decode error or an unexpected empty response.
* The proxy server now applies error-limiting to the correct node when
handling a recoverable node error.
* Account, container, and object log fields are now correctly identified
when returning `BadDigest` responses to S3 requests.
* Previously, under some circumstances, a non-standard config option such
as `RECLAIM_AGE` might get parsed as `reclaim_age` for some processes
but ignored by others. Now, all config parsing is case-sensitive;
non-standard names will always be ignored.
* Reduced the backend load of making `?versions` requests to a container
that has never had object versioning enabled.
* Improved formatting of meta and sysmeta for `swift-account-info` and
`swift-container-info`.
* The standard-library `logging` module is no longer monkey-patched when
importing `swift.common.utils`, making it easier to re-use swift code
in other contexts.
* Removed the hard dependency on netifaces; it may still be used if the
`getifaddrs` C function is not available. This fallback support may be
removed in a future release.
* Various other minor bug fixes and improvements.
swift (2.31.1, OpenStack Antelope)
* Sharding fixes
* Shards no longer report stats to the root database when they are in
the CREATED state.
* Sharding metadata is no longer cleared when databases are deleted.
This could previously cause deleted shards that still had rows to
become stuck and never move them to the correct database.
* Fixed a performance regression in the handling of misplaced objects.
* Swift path and on-disk path are now included with all sharder logging.
* `s3token` no longer mangles request paths that include the Access Key ID.
* User metadata is now exposed via CORS when encryption is enabled,
matching the behavior when encryption is not enabled.
* Fewer backend requests are now required when account or container
information is missing from memcache.
* Fixed logging of IP and port in the proxy-server; in particular,
internal clients now correctly log about the replication IP/port.
* Fixed a bug in the object replicator that would cause an under-reporting
of failures.
* Various other minor bug fixes.
swift (2.31.0)
* S3 API improvements
* Fixed a security issue in how `s3api` handles XML parsing that allowed
authenticated S3 clients to read arbitrary files from proxy servers.
Refer to CVE-2022-47950 for more information.
* Fixed a server error when handling malformed CompleteMultipartUpload
requests.
* Improved error reporting when attempting to set invalid `X-Delete-At`
or `X-Delete-After` values via the S3 API.
* Sharding improvements
* Sync more shard ranges from the root database to the shards. This
helps ensure shard range repairs effected at the root make their way
to shards that would otherwise be stuck trying to further divide
into sub-shards.
* Added a `merge` subcommand to `swift-manage-shard-ranges` to merge
arbitrary shard ranges into a container DB. Minimal safety checks
are performed; it should only be used for emergency shard range
manipulation by expert users.
* Improved performance of `delimiter` listings for sharded containers.
* Added more safety checks to the `repair` subcommand of
`swift-manage-shard-ranges`.
* Better handle `EOFError` and `KeyboardInterrupt` when prompting for
input in `swift-manage-shard-ranges`.
* Warnings are now emitted when sharding appears to have become stuck.
Use the new `container_sharding_timeout` option to configure the
"stuck" threshold; the default is 48 hours.
* Stop warning about transient overlaps when auditing shard ranges.
* Metrics improvements
* Added timing stats for memcached operations.
* Renamed and improved the granularity of shard range cache and
backend stats. Metrics dashboards may need to be updated.
* Emit stats when backend nodes are error-limited.
* Added support for Python 3.10.
* Added an optional `backend_ratelimit` middleware for backend servers.
See the backend server sample configuration files for more information.
* Added the ability to configure a chance to skip checking memcache when
querying account and container information. This allows some fraction
of traffic to go to disk and refresh memcache before the key ages out.
Recommended values for the new `account_existence_skip_cache_pct` and
`container_existence_skip_cache_pct` options are in the range of
0.0 to 0.01.
* Static large object segments may now be deleted asynchronously by
default. Operators may return to the old behavior by disabling the
`allow_async_delete` option in the `[filter:slo]` section
in their proxy-server.conf.
* Absolute-form request targets are now accepted. This enables access for
certain clients and SDKs (including some older versions of rclone that
were using an old version of aws-sdk-go).
* Fixed a path-rewriting bug introduced in Python 3.7.14, 3.8.14, 3.9.14,
and 3.10.6 that could cause some `domain_remap` requests to be routed to
the wrong object.
* Fixed a server error when attempting to access data in a deleted
container that had an erasure-coded storage policy.
* Improved error messages to clients that encounter errors using the
`formpost` middleware.
* Removed some inappropriate error-suppression when locking account and
container databases.
* Improved server start-up time when using multiple workers.
* Removed some unnecessary locking when logging.
* Added some basic object-metadata validation; invalid diskfiles will be
quarantined via the auditor or reconstructor.
* Enhanced logging when error-limiting a backend node.
* Various other minor bug fixes and improvements.
swift (2.30.1, zed stable backports)
* Fixed a security issue in how `s3api` handles XML parsing that allowed
authenticated S3 clients to read arbitrary files from proxy servers.
Refer to CVE-2022-47950 for more information.
* Fixed a path-rewriting bug introduced in Python 3.7.14, 3.8.14, 3.9.14,
and 3.10.6 that could cause some `domain_remap` requests to be routed to
the wrong object.
swift (2.30.0, OpenStack Zed)
* Sharding improvements
* The `swift-manage-shard-ranges` tool has a new mode to repair gaps
in the namespace.
* Misplaced tombstone records are now properly cleaved.
* Fixed a bug where the sharder could fail to find a device to use for
cleaving.
* Databases marked deleted are now processed by the sharder.
* More information is now synced to the fresh database when sharding.
Previously, a database could lose the fact that it had been marked
as deleted.
* Shard ranges with no rows to cleave could previously be left in the
CREATED state after cleaving. Now, they are advanced to CLEAVED.
* Metrics are now emitted for whether databases used for cleaving
were created or already existed, allowing a better understanding
of the reason for handoffs in the cluster.
* Misplaced-record stats are now also emitted to statsd. Previously,
these were only available in logs.
* S3 API improvements
* Constant-time string comparisons are now used when checking signatures.
* Fixed cross-policy object copies. Previously, copied data would
always be written using the source container's policy. Now, the
destination container's policy will be used, avoiding availability
issues and unnecessary container-reconciler work.
* More headers are now copied from multi-part upload markers to their
completed objects, including `Content-Encoding`.
* When running with `s3_acl` disabled, `bucket-owner-full-control` and
`bucket-owner-read` canned ACLs will be translated to the same Swift
ACLs as `private`.
* The S3 ACL and Delete Multiple APIs are now less case-sensitive.
* Improved the error message when deleting a bucket that's ever had
versioning enabled and still has versions in it.
* `LastModified` timestamps in listings are now rounded up to whole
seconds, like they are in responses from AWS.
* Proxy logging for Complete Multipart Upload requests is now more
consistent when requests have been retried.
* Logging improvements
* Signal handling is more consistently logged at notice level.
Previously, signal handling would sometimes be logged at info
or error levels.
* The message template for proxy logging may now include a
`{domain}` field for the client-provided `Host` header.
* The object-replicator now logs successful rsync transfers at debug
instead of info.
* Added a `log_rsync_transfers` option to the object-replicator.
Set it to false to disable logging rsync "send" lines; during
large rebalances, such logging can overwhelm log aggregation
while providing little useful information.
* Transaction IDs are now only included in daemon log lines
in a request/response context.
* Fixed a socket leak when clients try to delete a non-SLO as though
it were a Static Large Object.
* The formpost digest algorithm is now configurable via the new
`allowed_digests` option, and support is added for both SHA-256
and SHA-512. Supported formpost digests are exposed to clients in
`/info`. Additionally, formpost signatures can now be base64 encoded.
* Added metrics to the formpost and tempurl middlewares to monitor
digest usage in signatures.
* SHA-1 signatures are now deprecated for the formpost and tempurl
middlewares. At some point in the future, SHA-1 will no longer be
enabled by default; eventually, support for it will be removed
entirely.
* Improved compatibility with certain FIPS-mode-enabled systems.
* Added a `ring_ip` option for various object services. This may be
used to find own devices in the ring in a containerized environment
where the `bind_ip` may not appear in the ring at all.
* Account and container replicators can now be configured with a
`handoff_delete` option, similar to object replicators and
reconstructors. See the sample config for more information.
* Developers using Swift's memcache client may now opt in to having
a `MemcacheConnectionError` be raised when no connection succeeded
using a new `raise_on_error` keyword argument to `get`/`set`.
* The tempurl middleware has been updated to return a 503 if storing a
token in memcache fails. Third party authentication middlewares are
encouraged to also use the new `raise_on_error` keyword argument
when storing ephemeral tokens in memcache.
* Pickle support has been removed from Swift's memcache client. Support
had been deprecated since Swift 1.7.0.
* Device names are now included in new database IDs. This provides more
context when examining incoming/outgoing sync tables or sharding
CleaveContexts.
* Database replication connections are now closed following an error
or timeout. This prevents a traceback in some cases when the replicator
tries to reuse the connection.
* `ENOENT` and `ENODATA` errors are better handled in the object
replicator and auditor.
* Improved object update throughput by shifting some shard range
filtering from Python to SQL.
* Include `Vary: Origin` header when CORS responses vary by origin.
* The staticweb middleware now allows empty listings at the root of
a container. Previously, this would result in a 404 response.
* Ring builder output tables better display weights over 1000.
* Various other minor bug fixes and improvements.
swift (2.29.2, yoga stable backports)
* Fixed a security issue in how `s3api` handles XML parsing that allowed
authenticated S3 clients to read arbitrary files from proxy servers.
Refer to CVE-2022-47950 for more information.
* Constant-time string comparisons are now used when checking S3 API
signatures.
* Fixed a path-rewriting bug introduced in Python 3.7.14, 3.8.14, 3.9.14,
and 3.10.6 that could cause some `domain_remap` requests to be routed to
the wrong object.
* Improved compatibility with certain FIPS-mode-enabled systems.
swift (2.29.1, OpenStack Yoga)
* This is the final stable branch that will support Python 2.7.
* Fixed s3v4 signature calculation when the client sends an un-encoded
path in the request.
* Fixed multiple issues in s3api involving Multipart Uploads with
non-ASCII names.
* The object-updater now defers rate-limited updates to the end of its
cycle; these deferred updates will be processed (at the limited rate)
until the configured `interval` elapses. A new `max_deferred_updates`
option may be used to bound the deferral queue.
* Empty account and container partition directories are now cleaned up
immediately after replication, rather than needing to wait for an
additional replication cycle.
* The object-expirer now only cleans up empty containers. Previously, it
would attempt to delete all processed containers, regardless of whether
there were entries which were skipped or had errors.
* A new `item_size_warning_threshold` option may be used to monitor for
values that are approaching the limit of what can be stored in memcache.
See the memcache sample config for more information.
* Internal clients now correctly use their configured User-Agent in
backend requests, rather than only using it for logging.
* Various other minor bug fixes and improvements.
swift (2.29.0)
* S3 API improvements
* CORS preflights are now allowed for pre-signed URLs.
* The `storage_domain` option now accepts a comma-separated list of
storage domains. This allows multiple storage domains to configured
for use with virtual-host style addressing.
* Fixed the types of configured values in /info response.
* Fixed a server error when trying to copy objects with non-ASCII names.
* Fixed a server error when uploading objects with very long names.
A KeyTooLongError is now returned.
* Fixed an error when multi-deleting MPUs when SLO async-deletes
are enabled.
* Fixed an error that allowed list-uploads and list-parts requests to
return incomplete or out-of-order results.
* Fixed several bugs when dealing with non-ASCII object names and
multipart uploads.
* Reduced the overhead of retrieving bucket and object ACLs.
* Replication, reconstruction, and diskfile improvements
* The reconstructor now uses the replication network to fetch fragments
for reconstruction.
* Added the ability to limit how many objects per handoff partition
will be reverted in a reconstructor cycle using the new
`max_objects_per_revert` option. This may be useful to reduce
ssync timeouts and lock contention, ensuring that progress is made
during rebalances.
* Ensure that non-durable data and .meta files are purged from handoffs
after syncing.
* Fixed tracebacks when there's a race to mark a file durable or delete it.
* Improved cooperative multitasking during ssync.
* Upon detecting a ring change, the reconstructor now only aborts the
jobs for that ring and continues processing jobs for other rings.
* Fixed a traceback when logging about a lock timeout in the replicator.
* Object updater improvements
* Added the ability to ratelimit updates (approximately) per-container
using the new `max_objects_per_container_per_second` option. This may
be used to limit requests to already-overloaded containers while still
making progress on updates to other containers.
* Added timing stats by response code.
* Updates are now sent over the replication network.
* Fixed a race condition where swift would attempt to quarantine
recently-deleted updates.
* Memcache improvements
* Added the ability to configure a chance to skip checking memcache when
querying shard ranges. This allows some fraction of traffic to go to
disk and refresh memcache before the key ages out. Recommended values
for the new `container_updating_shard_ranges_skip_cache_pct` and
`container_listing_shard_ranges_skip_cache_pct` options are in the
range of 0.0 to 0.1.
* Added stats for shard range cache hits, misses, and skips.
* Improved handling of timeouts and other errors when obtaining a
connection to memcached.
* Recon improvements
* Added object-reconstructor stats to recon.
* Each object-server IP is now queried only once when reporting disk
usage. Previously, each port in the ring would be queried; when using
servers-per-port, this could dramatically overstate the disk capacity
in the cluster.
* Fixed a security issue where tempurl and s3api signatures were logged in
full. This allowed an attacker with access to log data to perform replay
attacks, potentially accessing or overwriting cluster data. Now, such
signatures are redacted in a manner similar to auth tokens; see the
`reveal_sensitive_prefix` option in `proxy-server.conf`.
See CVE-2017-8761 for more information.
* Added a new `swift.common.registry` module. This includes helper
functions `register_sensitive_header` and `register_sensitive_param`
which third party middleware authors may use to flag headers and query
parameters for redaction when logging. For more information, see
https://docs.openstack.org/swift/latest/misc.html#module-swift.common.registry
* Added the ability to configure project-scope read-only roles for
keystoneauth using the new `project_reader_roles` option.
* The cname_lookup middleware now works with dnspython 2.0 and later.
* The internal clients used by the container-reconciler, container-sharder,
container-sync, and object-expirer daemons now use a more-descriptive
`<daemon>-ic` log name, rather than `swift`. If you previously
configured the `log_name` option in `internal-client.conf`, you must
now use the `set log_name = <value>` syntax to configure it, even if
no value is set in the `[DEFAULT]` section. This may be done prior to
upgrading.
* Fixed a bug that allowed some statsd metrics to be annotated with the
wrong backend layer.
* The `StatsdClient.set_prefix` method is now deprecated and
may be removed in a future release; by extension, so is the
`LogAdapter.set_statsd_prefix` method. Middleware developers should
use the `statsd_tail_prefix` argument to `get_logger` instead.
* Fixed a traceback in the account-server when there's no account
database on disk to receive a container update. The account-server
now correctly 404s.
* The container-updater will quarantine container databases if all
replicas for the account respond 404.
* Fixed a proxy-server error when the read-only middleware tried to
handle non-Swift paths (such as may be used by third-party middleware).
* Some client behaviors that the proxy previously logged at warning have
been lowered to info.
* Removed translations from most logging.
* Various other minor bug fixes and improvements.
swift (2.28.1, xena stable backports)
* Fixed a security issue in how `s3api` handles XML parsing that allowed
authenticated S3 clients to read arbitrary files from proxy servers.
Refer to CVE-2022-47950 for more information.
* Constant-time string comparisons are now used when checking S3 API
signatures.
* Fixed a path-rewriting bug introduced in Python 3.7.14, 3.8.14, 3.9.14,
and 3.10.6 that could cause some `domain_remap` requests to be routed to
the wrong object.
* Improved compatibility with certain FIPS-mode-enabled systems.
* Ensure that non-durable data and .meta files are purged from handoffs
after syncing.
swift (2.28.0, OpenStack Xena)
* Sharding improvements:
* When building a listing from shards, any failure to retrieve
listings will result in a 503 response. Previously, failures
fetching a partiucular shard would result in a gap in listings.
* Container-server logs now include the shard path in the referer
field when receiving stat updates.
* Added a new config option, `rows_per_shard`, to specify how many
objects should be in each shard when scanning for ranges. The default
is `shard_container_threshold / 2`, preserving existing behavior.
* Added a new config option, `minimum_shard_size`. When scanning
for shard ranges, if the final shard would otherwise contain
fewer than this many objects, the previous shard will instead
be expanded to the end of the namespace (and so may contain up
to `rows_per_shard + minimum_shard_size` objects). This reduces
the number of small shards generated. The default value is
`rows_per_shard / 5`.
* Added a new config option, `shrink_threshold`, to specify the
absolute size below which a shard will be considered for shrinking.
This overrides the `shard_shrink_point` configuration option, which
expressed this as a percentage of `shard_container_threshold`.
`shard_shrink_point` is now deprecated.
* Similar to above, `expansion_limit` was added as an absolute-size
replacement for the now-deprecated `shard_shrink_merge_point`
configuration option.
* The sharder now correctly identifies and fails audits for shard
ranges that overlap exactly.
* The sharder and swift-manage-shard-ranges now consider total row
count (instead of just object count) when deciding whether a shard
is a candidate for shrinking.
* If the sharder encounters shard range gaps while cleaving, it will
now log an error and halt sharding progress. Previously, rows may
not have been moved properly, leading to data loss.
* Sharding cycle time and last-completion time are now available via
swift-recon.
* Fixed an issue where resolving overlapping shard ranges via shrinking
could prematurely mark created or cleaved shards as active.
* `swift-manage-shard-ranges` improvements:
* Exit codes are now applied more consistently:
- 0 for success
- 1 for an unexpected outcome
- 2 for invalid options
- 3 for user exit
As a result, some errors that previously resulted in exit code 2
will now exit with code 1.
* Added a new 'repair' command to automatically identify and
optionally resolve overlapping shard ranges.
* Added a new 'analyze' command to automatically identify overlapping
shard ranges and recommend a resolution based on a JSON listing
of shard ranges such as produced by the 'show' command.
* Added a `--includes` option for the 'show' command to only output
shard ranges that may include a given object name.
* Added a `--dry-run` option for the 'compact' command.
* The 'compact' command now outputs the total number of compactible
sequences.
* S3 API improvements:
* Added an option, `ratelimit_as_client_error`, to return 429s for
rate-limited responses. Several clients/SDKs have seem to support
retries with backoffs on 429, and having it as a client error
cleans up logging and metrics. By default, Swift will respond 503,
matching AWS documentation.
* Fixed a server error in bucket listings when `s3_acl` is enabled
and staticweb is configured for the container.
* Fixed a server error when a client exceeds `client_timeout` during an
upload. Now, a `RequestTimeout` error is correctly returned.
* Fixed a server error when downloading multipart uploads/static large
objects that have missing or inaccessible segments. This is a state
that cannot arise in AWS, so a new `BrokenMPU` error is returned,
indicating that retrying the request is unlikely to succeed.
* Fixed several issues with the prefix, marker, and delimiter
parameters that would be mirrored back to clients when listing
buckets.
* Partition power increase improvements:
* The relinker now spawns multiple subprocesses to process disks
in parallel. By default, one worker is spawned per disk; use the
new `--workers` option to control how many subprocesses are used.
Use `--workers=0` to maintain the previous behavior.
* The relinker now performs eventlet-hub selection the same way as
other daemons. In particular, `epolls` will no longer be selected,
as it seemed to cause occassional hangs.
* The relinker can now target specific storage policies or
partitions by using the new `--policy` and `--partition`
options.
* Partitions that encountered errors during relinking are no longer
marked as completed in the relinker state file. This ensures that
a subsequent relink will retry the failed partitions.
* Partition cleanup is more robust, decreasing the likelihood of
leaving behind mostly-empty partitions from the old partition
power.
* Improved relinker progress logging, and started collecting
progress information for swift-recon.
* Cleanup is more robust to files and directories being deleted by
another process.
* The relinker better handles data found from earlier partition power
increases.
* The relinker better handles tombstones found for the same object
but with different inodes.
* The reconciler now defers working on policies that have a partition
power increase in progress to avoid issues with concurrent writes.
* Erasure coding fixes:
* Added the ability to quarantine EC fragments that have no (or few)
other fragments in the cluster. A new configuration option,
`quarantine_threshold`, in the reconstructor controls the point at
the fragment will be quarantined; the default (0) will never
quarantine. Only fragments older than `quarantine_age` (default:
`reclaim_age`) may be quarantined. Before quarantining, the
reconstructor will attempt to fetch fragments from handoff nodes
in addition to the usual primary nodes; a new `request_node_count`
option (default `2 * replicas`) limits the total number of nodes to
contact.
* Added a delay before deleting non-durable data. A new configuration
option, `commit_window` in the `[DEFAULT]` section of
object-server.conf, adjusts this delay; the default is 60 seconds. This
improves the durability of both back-dated PUTs (from the reconciler or
container-sync, for example) and fresh writes to handoffs by preventing
the reconstructor from deleting data that the object-server was still
writing.
* Improved proxy-server and object-reconstructor logging when data
cannot be reconstructed.
* Fixed an issue where some but not all fragments having metadata
applied could prevent reconstruction of missing fragments.
* Server-side copying of erasure-coded data to a replicated policy no
longer copies EC sysmeta. The previous behavior had no material
effect, but could confuse operators examining data on disk.
* Python 3 fixes:
* Fixed a server error when performing a PUT authorized via
tempurl with some proxy pipelines.
* Fixed a server error during GET of a symlink with some proxy
pipelines.
* Fixed an issue with logging setup when /dev/log doesn't exist
or is not a UNIX socket.
* The container-reconciler now scales out better with new `processes`,
`process`, and `concurrency` options, similar to the object-expirer.
* The dark-data audit watcher now skips objects younger than a new
configurable `grace_age` period. This avoids issues where data
could be flagged, quarantined, or deleted because of listing
consistency issues. The default is one week.
* The dark-data audit watcher now requires that all primary locations
for an object's container agree that the data does not appear in
listings to consider data "dark". Previously, a network partition
that left an object node isolated could cause it to quarantine or
delete all of its data.
* More daemons now support systemd notify sockets.
* `EPIPE` errors no longer log tracebacks.
* The account and container auditors now log and update recon before
going to sleep.
* The object-expirer logs fewer client disconnects.
* `swift-recon-cron` now includes the last time it was run in the recon
information.
* `EIO` errors during read now cause object diskfiles to be quarantined.
* The formpost middleware now properly supports uploading multiple files
with different content-types.
* Various other minor bug fixes and improvements.
swift (2.27.0, OpenStack Wallaby)
* Added "audit watcher" hooks to allow operators to run arbitrary code
against every diskfile in a cluster. For more information, see
https://docs.openstack.org/swift/latest/development_watchers.html
* Added support for system-scoped "reader" roles when authenticating using
Keystone. Operators may configure this using the `system_reader_roles`
option in the `[filter:keystoneauth]` section of their proxy-server.conf.
A comparable group, `.reseller_reader`, is now available for development
purposes when authenticating using tempauth.
* Allow static large object segments to be deleted asynchronously.
Operators may opt into this new behavior by enabling the new
`allow_async_delete` option in the `[filter:slo]` section
in their proxy-server.conf. For more information, see
https://docs.openstack.org/swift/latest/overview_large_objects.html#deleting-a-large-object
* Added the ability to connect to memcached over TLS. See the
`tls_*` options in etc/memcache.conf-sample
* The proxy-server now caches 'listing' shards, improving listing
performance for sharded containers. A new config option,
`recheck_listing_shard_ranges`, controls the cache time and defaults to
10 minutes; set it to 0 to disable caching (the previous behavior).
* Added a new optional proxy-logging field `{wire_status_int}` for the
status code returned to the client. For more information, see
https://docs.openstack.org/swift/latest/logs.html#proxy-logs
* Errors downloading a Static Large Object that cause a shorter-than-expected
response are now logged as 500s.
* Memcache client error-limiting is now configurable. See the
`error_suppression_*` options in etc/memcache.conf-sample
* Added `tasks_per_second` option to rate-limit the object-expirer.
* Added `usedforsecurity` annotations for use on FIPS-compliant systems.
* Added an option to write EC fragments with legacy CRC to ensure a smooth
upgrade from liberasurecode<=1.5.0 to >=1.6.2. For more information, see
https://bugs.launchpad.net/liberasurecode/+bug/1886088
* **Known Issue**: Operators should verify that encryption is not enabled
in their reconciler pipelines; having it enabled there may harm data
durability. For more information, see https://launchpad.net/bugs/1910804
* S3 API improvements:
* Fixed a bug that prevented the s3api pipeline validation described in
proxy-server.conf-sample from being performed. As documented, operators
can disable this via the `auth_pipeline_check` option if proxy startup
fails with validation errors.
* Make allowable clock skew configurable, with a default value of
15 minutes to match AWS. Note that this was previously hardcoded at
5 minutes; operators may want to preserve the prior behavior by setting
`allowable_clock_skew = 300` in the `[filter:s3api]` section of their
proxy-server.conf.
* Fixed an issue where SHA mismatches in client XML payloads would cause
a server error. Swift now correctly responds with a client error about
the bad digest.
* Fixed an issue where non-base64 signatures would cause a server error.
Swift now correctly responds with a client error about the invalid
digest.
* Container ACLs are now cloned to the `+segments` container when it is
created.
* The correct storage policy is now logged for S3 requests.
* Added the ability to configure auth region in s3token middleware.
* CORS-related headers are now passed through appropriately when using
the S3 API. Note that allowed origins and other container metadata
must still be configured through the Swift API as documented at
https://docs.openstack.org/swift/latest/cors.html
Preflight requests do not contain enough information to map a
bucket to an account/container pair; a new cluster-wide option
`cors_preflight_allow_origin` may be configured for such OPTIONS
requests. The default (blank) rejects all S3 preflight requests.
* Sharding improvements:
* Prevent shard databases from losing track of their root database when
deleted.
* Prevent sharded root databases from being reclaimed to ensure that
shards can detect that they have been deleted.
* A `--no-auto-shard` option has been added to `swift-container-sharder`.
* The sharder daemon has been enhanced to better support the shrinking
of shards that are no longer required. Shard containers will now
discover from their root container if they should be shrinking. They
will also discover the shards into which they should shrink, which may
include the root container itself.
* A 'compact' command has been added to `swift-manage-shard-ranges` that
enables sequences of contiguous shards with low object counts to be
compacted into another existing shard, or into the root container.
* `swift-manage-shard-ranges` can now accept a config file; this
may be used to ensure consistency of threshold values with the
container-sharder config.
* Overlapping shrinking shards no longer generate audit warnings; these
are expected to sometimes overlap.
* The sharding progress reports in recon cache now continue to be included
for a period of time after sharding has completed. The time period
may be configured using the `recon_sharded_timeout` option in the
`[container-sharder]` section of container-server.conf, and defaults
to 12 hours.
* Add root containers with compactible ranges to recon cache.
* Expose sharding statistics in the backend recon middleware.
* Replication improvements:
* Fixed a race condition in ssync that could lead to a loss of data
durability (or even loss of data, for two-replica policies) when some
object servers have outdated rings. Replication via rsync is likely
still affected by a similar bug.
* Non-durable fragments can now be reverted from handoffs.
* The post-rsync REPLICATE call no longer recalculates hashes immediately.
* Hashes are no longer invalidated after a successful ssync; they were
already invalidated during the data transfer.
* Reduced log noise for common ssync errors.
* Python 3 fixes:
* Added support for Python 3.9.
* Staticweb correctly handles listings when paths include non-ASCII
characters.
* S3 API now allows multipart uploads with non-ASCII characters in the
object name.
* Fixed an import-ordering issue in `swift-dispersion-populate`.
* Partition power increase improvements:
* Fixed a bug where stale state files would cause misplaced data during
multiple partition power increases.
* Removed a race condition that could cause newly-written data to not be
linked into the new partition for the new partition power.
* Improved safety during cleanup to ensure files have been relinked
appropriately before unlinking.
* Added an option to drop privileges when running the relinker as root.
* Added an option to rate-limit how quickly data files are relinked or
cleaned up. This may be used to reduce I/O load during partition power
increases, improving end-user performance.
* Rehash partitions during the partition power increase. Previously, we
relied on the replication engine to perform the rehash, which could
cause an unexpected I/O spike after a partition power increase.
* Warn when relinking/cleaning up and any disks are unmounted.
* Log progress per partition when relinking/cleaning up.
* During clean-up, stop warning about tombstones that got reaped from
the new location but not the old.
* Added the ability to read options from object-server.conf, similar to
background daemons.
* Turned off thread-logging when monkey-patching with eventlet. This
addresses a potential hang in the proxy-server while logging client
disconnects.
* Fixed a bug that could cause EC GET responses to return a server error.
* Fixed an issue with `swift-drive-audit` when run around New Year's.
* Server errors encountered when validating the first segment of a Static or
Dynamic Large Object now return a 503 to the client, rather than a 409.
* Errors when setting keys in memcached are now logged. This helps
operators detect when shard ranges for caching have gotten too large to
be stored, for example.
* Various other minor bug fixes and improvements.
swift (2.26.0, OpenStack Victoria)
* Extend concurrent reads to erasure coded policies. Previously, the
options `concurrent_gets` and `concurrency_timeout` only applied to
replicated policies.
* Add a new `concurrent_ec_extra_requests` option to allow the proxy to
make some extra backend requests immediately. The proxy will respond as
soon as there are enough responses available to reconstruct.
* The concurrent read options (`concurrent_gets`, `concurrency_timeout`,
and `concurrent_ec_extra_requests`) may now be configured per
storage-policy.
* Replication servers can now handle all request methods. This allows
ssync to work with a separate replication network.
* All background daemons now use the replication network. This allows
better isolation between external, client-facing traffic and internal,
background traffic. Note that during a rolling upgrade, replication
servers may respond with `405 Method Not Allowed`. To avoid this,
operators should remove the config option `replication_server = true`
from their replication servers; this will allow them to handle all
request methods before upgrading.
* S3 API improvements:
* Fixed some SignatureDoesNotMatch errors when using the AWS .NET SDK.
* Add basic read support for object tagging. This improves
compatibility with AWS CLI version 2. Write support is not
yet implemented, so the tag set will always be empty.
* CompleteMultipartUpload requests may now be safely retried.
* Improved quota-exceeded error messages.
* Improved logging and statsd metrics. Be aware that this will cause
an increase in the proxy-logging statsd metrics emited for S3
responses. However, this should more accurately reflect the state
of the system.
* S3 requests are now less demanding on the container layer.
* Python 3 bug fixes:
* Fixed an error when reading encrypted data that was written while
running Python 2 for a path that includes non-ASCII characters. This
was caused by a difference in string types that resulted in
ambiguity when decrypting. To prevent the ambiguity for new data, set
`meta_version_to_write = 3` in your keymaster configuration after
upgrading all proxy servers.
If upgrading from Swift 2.20.0 or Swift 2.19.1 or earlier, set
`meta_version_to_write = 1` in your keymaster configuration prior
to upgrading.
* Object expiration respects the `expiring_objects_container_divisor`
config option.
* `fallocate_reserve` may be specified as a percentage in more places.
* The ETag-quoting middleware no longer raises TypeErrors.
* Sharding improvements:
* Prevent object updates from auto-creating shard containers. This
ensures more consistent listings for sharded containers during
rebalances.
* Deleted shard containers are no longer considered root containers.
This prevents unnecessary sharding audit failures and allows the
deleted shard database to actually be unlinked.
* `swift-container-info` now summarizes shard range information.
Pass `-v`/`--verbose` if you want to see all of them.
* Improved container-sharder stat reporting to reduce load on root
container databases.
* Don't inject shard ranges when user quits.
* Servers now open one listen socket per worker, ensuring each worker
serves roughly the same number of concurrent connections.
* Server workers may now be gracefully terminated via `SIGHUP` or
`SIGUSR1`. The parent process will then spawn a fresh worker.
* During rebalances, clients should no longer get 404s for data that
exists but whose replicas are overloaded.
* Improved cache management for account and container responses.
* Allow proxy-logging middlewares to be configured more independently.
* Allow operators to pass either raw or URL-quoted paths to
swift-get-nodes. Notably, this allows swift-get-nodes to work with
the reserved namespace used for object versioning.
* Container read ACLs now work with object versioning. This only
allows access to the most-recent version via an unversioned URL.
* Improved how containers reclaim deleted rows to reduce locking and object
update throughput.
* Large object reads log fewer client disconnects.
* Allow ratelimit to be placed multiple times in a proxy pipeline,
such as both before s3api and auth (to handle swift requests without
needing to make an auth decision) and after (to limit S3 requests).
* Shuffle object-updater work. This somewhat reduces the impact a
single overloaded database has on other containers' listings.
* Fix a proxy-server error when retrieving erasure coded data when
there are durable fragments but not enough to reconstruct.
* Fix an error in the proxy server when finalizing data.
* Improve performance when increasing partition power.
* Various other minor bug fixes and improvements.
swift (2.25.1, ussuri stable backports)
* Python 3 bug fixes:
* Fixed an error when reading encrypted data that was written while
running Python 2 for a path that includes non-ASCII characters. This
was caused by a difference in string types that resulted in
ambiguity when decrypting. To prevent the ambiguity for new data, set
`meta_version_to_write = 3` in your keymaster configuration after
upgrading all proxy servers.
If upgrading from Swift 2.20.0 or Swift 2.19.1 or earlier, set
`meta_version_to_write = 1` in your keymaster configuration prior
to upgrading.
* Object expiration respects the `expiring_objects_container_divisor`
config option.
* `fallocate_reserve` may be specified as a percentage in more places.
* The ETag-quoting middleware no longer raises TypeErrors.
* Improved how containers reclaim deleted rows to reduce locking and object
update throughput.
* Fix a proxy-server error when retrieving erasure coded data when
there are durable fragments but not enough to reconstruct.
* Fixed some SignatureDoesNotMatch errors when using the AWS .NET SDK.
* Region name config option is now respected when configuring S3 credential
caching.
swift (2.25.0, OpenStack Ussuri)
* WSGI server processes can now notify systemd when they are ready.
* Added `ttfb` (Time to First Byte) and `pid` (Process ID) to the set
of available proxy-server log fields. For more information, see
https://docs.openstack.org/swift/latest/logs.html
* Improved proxy-server performance by reducing unnecessary locking,
memory copies, and eventlet scheduling.
* Reduced object-replicator and object-reconstructor CPU usage by only
checking that the device list is current when rings change.
* Improved performance of sharded container listings when performing
prefix listings.
* Improved container-sync performance when data has already been
deleted or overwritten.
* Account quotas are now enforced even on empty accounts.
* Getting an SLO manifest with `?format=raw` now responds with an ETag
that matches the MD5 of the generated body rather than the MD5 of
the manifest stored on disk.
* Provide useful status codes in logs for some versioning and symlink
subrequests that were previously logged as 499.
* Fixed 500 from cname_lookup middleware. Previously, if the looked-up
domain was used by domain_remap to update the request path, the
server would respond Internal Error.
* On Python 3, fixed an issue when reading or writing objects with a
content-type like `message/*`. Previously, Swift would fail to respond.
* On Python 3, fixed a RecursionError in swift-dispersion-report when
using TLS.
* Fixed a bug in the new object versioning API that would cause more
than `limit` results to be returned when listing.
* Various other minor bug fixes and improvements.
swift (2.24.0)
* Added a new object versioning mode, with APIs for querying and
accessing old versions. For more information, see the documentation
at https://docs.openstack.org/swift/latest/middleware.html#module-swift.common.middleware.versioned_writes.object_versioning
* Added support for S3 versioning using the above new mode.
* Added a new middleware to allow accounts and containers to opt-in to
RFC-compliant ETags. This may be useful when using Swift as an origin
for some content delivery networks. For more information, see the
documentation at https://docs.openstack.org/swift/latest/middleware.html#module-swift.common.middleware.etag_quoter
Clients should be aware of the fact that ETags may be quoted for RFC
compliance; this may become the default behavior in some future release.
* Proxy, account, container, and object servers now support "seamless
reloads" via `SIGUSR1`. This is similar to the existing graceful
restarts but keeps the server socket open the whole time, reducing
service downtime.
* New buckets created via the S3 API will now store multi-part upload
data in the same storage policy as other data rather than the
cluster's default storage policy.
* Device region and zone can now be changed via `swift-ring-builder`.
Note that this may cause a lot of data movement on the next rebalance
as the builder tries to reach full dispersion.
* Added support for Python 3.8.
* The container sharder can now handle containers with special
characters in their names.
* Internal client no longer logs object DELETEs as status 499.
* Objects with an `X-Delete-At` value in the far future no longer cause
backend server errors.
* The bulk extract middleware once again allows clients to specify metadata
(including expiration timestamps) for all objects in the archive.
* Container sync now synchronizes static symlinks in a way similar to
static large objects.
* `swift_source` is set for more sub-requests in the proxy-server. See
https://docs.openstack.org/swift/latest/logs.html#swift-source
* Errors encountered while validating static symlink targets no longer
cause BadResponseLength errors in the proxy-server.
* On Python 3, the KMS keymaster now works with secrets stored
in Barbican with a text/plain payload-content-type.
* On Python 3, the formpost middleware now works with unicode file names.
* Several utility scripts now work better on Python 3:
* swift-account-audit
* swift-dispersion-populate
* swift-drive-recon
* swift-recon
* On Python 3, certain S3 API headers are now lower case as they
would be coming from AWS.
* Per-service `auto_create_account_prefix` settings are now deprecated
and may be ignored in a future release; if you need to use this, please
set it in the `[swift-constraints]` section of /etc/swift/swift.conf.
* Various other minor bug fixes and improvements.
swift (2.23.3, train stable backports)
* Sharding improvements:
* Prevent object updates from auto-creating shard containers. This
ensures more consistent listings for sharded containers during
rebalances.
* Deleted shard containers are no longer considered root containers.
This prevents unnecessary sharding audit failures and allows the
deleted shard database to actually be unlinked.
* The sharder daemon has been enhanced to better support the shrinking
of shards that are no longer required. Shard containers will now
discover from their root container if they should be shrinking. They
will also discover the shards into which they should shrink, which may
include the root container itself.
* Improved performance of sharded container listings when performing
prefix listings.
* Improved container-sharder stat reporting to reduce load on root
container databases.
* The container sharder can now handle containers with special
characters in their names.
* `swift-container-info` now summarizes shard range information.
Pass `-v`/`--verbose` if you want to see all of them.
* Don't inject shard ranges when user quits.
* Various other minor bug fixes and improvements.
* Python 3 bug fixes:
* Fixed a potential server error when uploading data via a tempurl.
* Fixed a potential server error when getting symlink details.
* Added the ability to connect to memcached over TLS. See the
`tls_*` options in etc/memcache.conf-sample
swift (2.23.2, train stable backports)
* Python 3 bug fixes:
* Fixed an error when reading encrypted data that was written while
running Python 2 for a path that includes non-ASCII characters. This
was caused by a difference in string types that resulted in
ambiguity when decrypting. To prevent the ambiguity for new data, set
`meta_version_to_write = 3` in your keymaster configuration after
upgrading all proxy servers.
If upgrading from Swift 2.20.0 or Swift 2.19.1 or earlier, set
`meta_version_to_write = 1` in your keymaster configuration prior
to upgrading.
* Fixed an issue when reading or writing objects with a content-type
like `message/*`. Previously, Swift would fail to respond.
* Object expiration respects the `expiring_objects_container_divisor`
config option.
* `fallocate_reserve` may be specified as a percentage in more places.
* The formpost middleware now works with unicode file names.
* Certain S3 API headers are now lower case as they would be coming
from AWS.
* Improved how containers reclaim deleted rows to reduce locking and object
update throughput.
* Fix a proxy-server error when retrieving erasure coded data when
there are durable fragments but not enough to reconstruct.
* Fixed 500 from cname_lookup middleware. Previously, if the looked-up
domain was used by domain_remap to update the request path, the
server would respond Internal Error.
* The bulk extract middleware once again allows clients to specify metadata
(including expiration timestamps) for all objects in the archive.
* Errors encountered while validating static symlink targets no longer
cause BadResponseLength errors in the proxy-server.
* Fixed some SignatureDoesNotMatch errors when using the AWS .NET SDK.
* Various other minor bug fixes and improvements.
swift (2.23.1, train stable backports)
* On Python 3, the KMS keymaster now works with secrets stored
in Barbican with a text/plain payload-content-type.
* Several utility scripts now work better on Python 3:
* swift-account-audit
* swift-dispersion-populate
* swift-drive-recon
* swift-recon
swift (2.23.0, OpenStack Train)
* Python 3.6 and 3.7 are now fully supported. Several py3-related
fixes are included:
* Removed a request-smuggling vector when running a mixed
py2/py3 cluster.
* Allow fallocate_reserve to be specified as a percentage.
* Fixed listings for sharded containers.
* Fixed non-ASCII account metadata handling.
* Fixed rsync output parsing.
* Fixed some title-casing of headers.
If you've been testing Swift on Python 3, upgrade at your earliest
convenience.
* Added "static symlinks", which perform some validation as they
follow redirects and include more information about their target
in container listings.
* Multi-character strings may now be used as delimiters in account
and container listings.
* Sharding improvements
* Container metadata related to sharding are now removed when no
longer needed.
* Empty container databases (such as might be created on handoffs)
now shard much more quickly.
* The proxy-server now ignores 404 responses from handoffs that have
no data when deciding on the correct response for object requests,
similar to what it already does for account and container requests.
* Static Large Object sizes in listings for versioned containers are
now more accurate.
* When refetching Static Large Object manifests, non-manifest responses
are now handled better.
* S3 API now translates 503 Service Unavailable responses to a more
S3-like response instead of raising an error.
* Improved proxy-to-backend requests to be more RFC-compliant.
* Dependency update: eventlet must be at least 0.25.0. This also
dragged forward minimum-supported versions of dnspython (1.15.0),
greenlet (0.3.2), and six (1.10.0).
* Various other minor bug fixes and improvements.
swift (2.22.0)
* Experimental support for Python 3.6 and 3.7 is now available.
Note that this requires eventlet>=0.25.0. All unit tests pass,
and running functional tests under Python 2 will pass against
services running under Python 3. Expect full support in the
next minor release.
* Log formats are now more configurable and include support for
anonymization. See the log_msg_template option in proxy-server.conf
and https://docs.openstack.org/swift/latest/logs.html#proxy-logs
for more information.
* Added an operator tool, swift-container-deleter, to asynchronously
delete some or all objects in a container using the object expirers.
* Swift-all-in-one Docker images are now built and published to
https://hub.docker.com/r/openstackswift/saio. These are intended
for use as development targets, but will hopefully be useful as a
starting point for other work involving containerizing Swift.
* The object-expirer may now be configured in object-server.conf.
This is in anticipation of a future change to allow the
object-expirer to be deployed on all nodes that run object-servers.
* Correctness improvements
* The proxy-server now ignores 404 responses from handoffs without
databases when deciding on the correct response for account and
container requests.
* Object writes to a container whose existence cannot be verified
now 503 instead of 404.
* Sharding improvements
* The container-replicator now only attempts to fetch shard ranges if
the remote indicates that it has shard ranges. Further, it does so
with a timeout to prevent the process from hanging in certain cases.
* The proxy-server now caches 'updating' shards, improving write
performance for sharded containers. A new config option,
`recheck_updating_shard_ranges`, controls the cache time; set it to
0 to disable caching.
* The container-replicator now correctly enqueues container-reconciler
work for sharded containers.
* S3 API improvements
* Unsigned payloads work with v4 signatures once more.
* Multipart upload parts may now be copied from other multipart uploads.
* CompleteMultipartUpload requests with a Content-MD5 now work.
* Content-Type can now be updated when copying an object.
* Fixed v1 listings that end with a non-ASCII object name.
* Background corruption-detection improvements
* Detect and remove invalid entries from hashes.pkl
* When object path is not a directory, just quarantine it,
rather than the whole suffix.
* Dependency updates: we've increased our minimum supported version
of cryptography to 2.0.2 and netifaces to 0.8. This is largely due
to the difficulty of continuing to test with the old versions.
If running Swift under Python 3, eventlet must be at least 0.25.0.
* Various other minor bug fixes and improvements.
swift (2.21.1, stein stable backports)
* Sharding improvements
* The container-replicator now only attempts to fetch shard ranges if
the remote indicates that it has shard ranges. Further, it does so
with a timeout to prevent the process from hanging in certain cases.
* The container-replicator now correctly enqueues container-reconciler
work for sharded containers.
* Container metadata related to sharding are now removed when no
longer needed.
* S3 API improvements
* Unsigned payloads work with v4 signatures once more.
* Multipart upload parts may now be copied from other multipart uploads.
* CompleteMultipartUpload requests with a Content-MD5 now work.
* Content-Type can now be updated when copying an object.
* Fixed v1 listings that end with a non-ASCII object name.
* Background corruption-detection improvements
* Detect and remove invalid entries from hashes.pkl
* When object path is not a directory, just quarantine it,
rather than the whole suffix.
* Static Large Object sizes in listings for versioned containers are
now more accurate.
* When refetching Static Large Object manifests, non-manifest responses
are now handled better.
* Cross-account symlinks now store correct account information in
container listings. This was previously fixed in 2.22.0.
* Requesting multiple ranges from a Dynamic Large Object now returns the
entire object instead of incorrect data. This was previously fixed in
2.23.0.
* When making backend requests, the proxy-server now ensures query
parameters are always properly quoted. Previously, the proxy would
encounter an error on Python 2.7.17 if the client included non-ASCII
query parameters in object requests. This was previously fixed in
2.23.0.
swift (2.21.0, OpenStack Stein)
* Change the behavior of the EC reconstructor to perform a
fragment rebuild to a handoff node when a primary peer responds
with 507 to the REPLICATE request. This changes EC to match the
existing behavior of replication when drives fail. After a
rebalance of EC rings (potentially removing unmounted/failed
devices), it's most IO efficient to run in handoffs_only mode to
avoid unnecessary rebuilds.
* O_TMPFILE support is now detected by attempting to use it
instead of looking at the kernel version. This allows older
kernels with backported patches to take advantage of the
O_TMPFILE functionality.
* Add slo_manifest_hook callback to allow other middlewares to
impose additional constraints on or make edits to SLO manifests
before being written. For example, a middleware could enforce
minimum segment size or insert data segments.
* Fixed an issue with multi-region EC policies that caused the EC
reconstructor to constantly attempt cross-region rebuild
traffic.
* Fixed an issue where S3 API v4 signatures would not be validated
against the body of the request, allowing a replay attack if
request headers were captured by a malicious third party.
* Display crypto data/metadata details in swift-object-info.
* formpost can now accept a content-encoding parameter.
* Fixed an issue where multipart uploads with the S3 API would
sometimes report an error despite all segments being upload
successfully.
* Multipart object segments are now actually deleted when the
multipart object is deleted via the S3 API.
* Swift now returns a 503 (instead of a 500) when an account
auto-create fails.
* Fixed a bug where encryption would store the incorrect key
metadata if the object name starts with a slash.
* Fixed an issue where an object server failure during a client
download could leave an open socket between the proxy and
client.
* Fixed an issue where deleted EC objects didn't have their
on-disk directories cleaned up. This would cause extra resource
usage on the object servers.
* Fixed issue where bulk requests using xml and expect
100-continue would return a malformed HTTP response.
* Various other minor bug fixes and improvements.
swift (2.20.0)
* S3 API compatibility updates
* Swift can now cache the S3 secret from Keystone to use for
subsequent requests. This functionality is disabled by default but
can be enabled by setting the `secret_cache_duration` in the s3token
section of the proxy server config to a number greater than 0.
* s3api now mimics the AWS S3 behavior of periodically sending
whitespace characters on a Complete Multipart Upload request to keep
the connection from timing out. Note that since a request could fail
after the initial 200 OK response has been sent, it is important to
check the response body to determine if the request succeeded.
* s3api now properly handles x-amz-metadata-directive headers on
COPY operations.
* s3api now uses concurrency (default 2) to handle multi-delete
requests. This allows multi-delete requests to be processed much
more quickly.
* s3api now mimics some forms of AWS server-side encryption
based on whether Swift's at-rest encryption functionality is enabled.
Note that S3 API users are now able to know more about how the
cluster is configured than they were previously, ie knowledge of
encryption at-rest functionality being enabled or not.
* s3api responses now include a '-' in multipart ETags.
For new multipart-uploads via the S3 API, the ETag that is
stored will be calculated in the same way that AWS uses. This
ETag will be used in GET/HEAD responses, bucket listings, and
conditional requests via the S3 API. Accessing the same object
via the Swift API will use the SLO Etag; however, in JSON
container listings the multipart upload etag will be exposed
in a new "s3_etag" key. Previously, some S3 clients would complain
about download corruption when the ETag did not have a '-'.
* S3 ETag for SLOs now include a '-'.
Ordinary objects in S3 use the MD5 of the object as the ETag,
just like Swift. Multipart Uploads follow a different format, notably
including a dash followed by the number of segments. To that end
(and for S3 API requests *only*), SLO responses via the S3 API have a
literal '-N' added on the end of the ETag.
* The default location is now set to "us-east-1". This is more likely
to be the default region that a client will try when using v4
signatures.
Deployers with clusters that relied on the old implicit default
location of "US" should explicitly set `location = US` in the
`[filter:s3api]` section of proxy-server.conf before upgrading.
* Add basic support for ?versions bucket listings. We still do not
have support for toggling S3 bucket versioning, but we can at least
support getting the latest versions of all objects.
* Fixed an issue with SSYNC requests to ensure that only one request
can be running on a partition at a time.
* Data encryption updates
* The kmip_keymaster middleware can now be configured directly in the
proxy-server config file. The existing behavior of using an external
config file is still supported.
* Multiple keymaster middlewares are now supported. This allows
migration from one key provider to another.
Note that secret_id values must remain unique across all keymasters
in a given pipeline. If they are not unique, the right-most keymaster
will take precedence.
When looking for the active root secret, only the right-most
keymaster is used.
* Prevent PyKMIP's kmip_protocol logger from logging at DEBUG.
Previously, some versions of PyKMIP would include all wire
data when the root logger was configured to log at DEBUG; this
could expose key material in logs. Only the kmip_keymaster was
affected.
* Fixed an issue where a failed drive could prevent the container sharder
from making progress.
* Storage policy definitions in swift.conf can now define the diskfile
to use to access objects. See the included swift.conf-sample file for
a description of usage.
* The EC reconstructor will now attempt to remove empty directories
immediately, while the inodes are still cached, rather than waiting
until the next run.
* Added a keep_idle config option to configure KEEPIDLE time for TCP
sockets. The default value is the old constant of 600.
* Add databases_per_second to the account-replicator,
container-replicator, and container-sharder. This prevents them from
using a full CPU core when they are not IO limited.
* Allow direct_client users to overwrite the X-Timestamp header.
* Various other minor bug fixes and improvements.
swift (2.19.2, rocky stable backports)
* Sharding improvements
* The container-replicator now only attempts to fetch shard ranges if
the remote indicates that it has shard ranges. Further, it does so
with a timeout to prevent the process from hanging in certain cases.
* The container-replicator now correctly enqueues container-reconciler
work for sharded containers.
* S3 API improvements
* Fixed an issue where v4 signatures would not be validated against
the body of the request, allowing a replay attack if request headers
were captured by a malicious third party. Note that unsigned payloads
still function normally.
* CompleteMultipartUpload requests with a Content-MD5 now work.
* Fixed v1 listings that end with a non-ASCII object name.
* Multipart object segments are now actually deleted when the
multipart object is deleted via the S3 API.
* Fixed an issue that caused Delete Multiple Objects requests with
large bodies to 400. This was previously fixed in 2.20.0.
* Fixed an issue where non-ASCII Keystone EC2 credentials would not get
mapped to the correct account. This was previously fixed in 2.20.0.
* Background corruption-detection improvements
* Detect and remove invalid entries from hashes.pkl
* When object path is not a directory, just quarantine it,
rather than the whole suffix.
* Fixed a bug where encryption would store the incorrect key
metadata if the object name starts with a slash.
* Fixed an issue where an object server failure during a client
download could leave an open socket between the proxy and
client.
* Static Large Object sizes in listings for versioned containers are
now more accurate.
* When refetching Static Large Object manifests, non-manifest responses
are now handled better.
* Cross-account symlinks now store correct account information in
container listings. This was previously fixed in 2.22.0.
* Requesting multiple ranges from a Dynamic Large Object now returns the
entire object instead of incorrect data. This was previously fixed in
2.23.0.
* When making backend requests, the proxy-server now ensures query
parameters are always properly quoted. Previously, the proxy would
encounter an error on Python 2.7.17 if the client included non-ASCII
query parameters in object requests. This was previously fixed in
2.23.0.
swift (2.19.1, rocky stable backports)
* Prevent PyKMIP's kmip_protocol logger from logging at DEBUG.
Previously, some versions of PyKMIP would include all wire
data when the root logger was configured to log at DEBUG; this
could expose key material in logs. Only the kmip_keymaster was
affected.
* Fixed an issue where a failed drive could prevent the container sharder
from making progress.
* Fixed a bug in how Swift uses eventlet that was exposed under high
concurrency.
swift (2.19.0, OpenStack Rocky)
* TempURLs now support IP range restrictions. Please see
https://docs.openstack.org/swift/latest/middleware.html#client-usage
for more information on how to use this additional restriction.
* Add support for multiple root encryption secrets for the trivial
and KMIP keymasters. This allows operators to rotate encryption
keys over time without needing to re-encrypt all existing data
in the cluster. Please see the included sample config files for
instructions on how to multiple encryption keys.
* The object updater now supports two configuration settings:
"concurrency" and "updater_workers". The latter controls how many
worker processes are spawned, while the former controls how many
concurrent container updates are performed by each worker
process. This should speed the processing of async_pendings.
On upgrade, a node configured with concurrency=N will still handle
async updates N-at-a-time, but will do so using only one process
instead of N.
If you have a config file like this:
[object-updater]
concurrency = <N>
and you want to take advantage of faster updates, then do this:
[object-updater]
concurrency = 8 # the default; you can omit this line
updater_workers = <N>
If you want updates to be processed exactly as before, do this:
[object-updater]
concurrency = 1
updater_workers = <N>
* When listing objects in a container in json format, static large
objects (SLOs) will now include an additional new "slo_etag" key
that matches the etag returned when requesting the SLO. The
existing "hash" key remains unchanged as the MD5 of the SLO
manifest. Text and XML listings are unaffected by this change.
* Log deprecation warnings for `run_pause`. This setting was
deprecated in Swift 2.4.0 and is replaced by `interval`.
It may be removed in a future release.
* Object reconstructor logs are now prefixed with information
about the specific worker process logging the message. This
makes reading the logs and understanding the messages much simpler.
* Lower bounds of dependencies have been updated to reflect what
is actually tested.
* SSYNC replication mode now removes as much of the directory
structure as possible as soon at it observes that the directory
is empty. This reduces the work needed for subsequent replication
passes.
* The container-updater now reports zero objects and bytes used for
child DBs in sharded containers. This prevents double-counting in
utilization reports.
* Add fallocate_reserve to account and container servers. This
allows disks shared between account/container and object rings to
avoid getting 100% full. The default value of 1% matches the
existing default on object servers.
* Added an experimental `swift-ring-composer` CLI tool to build
composite rings.
* Added an optional `read_only` middleware to make an entire cluster
or individual accounts read only.
* Fixed a bug where zero-byte PUTs would not work properly
with "If-None-Match: *" conditional requests.
* ACLs now work with unicode in user/account names.
* COPY now works with unicode account names.
* Improved S3 API compatibility.
* Lock timeouts in the container updater are now logged at INFO
level, not ERROR.
* Various other minor bug fixes and improvements.
swift (2.18.0)
* Added container sharding, an operator controlled feature that
may be used to shard very large container databases into a
number of smaller shard containers. This mitigates the issues
with one large DB by distributing the data across multiple
smaller databases throughout the cluster. Please read the full
overview at
https://docs.openstack.org/swift/latest/overview_container_sharding.html
* Provide an S3 API compatibility layer. The external "swift3"
project has been imported into Swift's codebase as the "s3api"
middleware.
* Added "emergency mode" hooks in the account and container replicators.
These options may be used to prioritize moving handoff
partitions to primary locations more quickly. This helps when
adding capacity to a ring.
- Added `-d <devs>` and `-p <partitions>` command line options.
- Added a handoffs-only mode.
* Add a multiprocess mode to the object replicator. Setting the
"replicator_workers" setting to a positive value N will result
in the replicator using up to N worker processes to perform
replication tasks. At most one worker per disk will be spawned.
Worker process logs will have a bit of information prepended so
operators can tell which messages came from which worker. The
prefix is "[worker M/N pid=P] ", where M is the worker's index,
N is the total number of workers, and P is the process ID. Every
message from the replicator's logger will have the prefix
* The object reconstructor will now fork all available worker
processes when operating on a subset of local devices.
* Add support for PROXY protocol v1 to the proxy server. This
allows the Swift proxy server to log accurate client IP
addresses when there is a proxy or SSL-terminator between the
client and the Swift proxy server. Example servers supporting
this PROXY protocol include stunnel, haproxy, hitch, and
varnish. See the sample proxy server config file for the
appropriate config setting to enable or disable this
functionality.
* In the ratelimit middleware, account whitelist and blacklist
settings have been deprecated and may be removed in a future
release. When found, a deprecation message will be logged.
Instead of these config file values, set X-Account-Sysmeta-
Global-Write-Ratelimit:WHITELIST and X-Account-Sysmeta-Global-
Write-Ratelimit:BLACKLIST on the particular accounts that need
to be whitelisted or blacklisted. System metadata cannot be added
or modified by standard clients. Use the internal client to set sysmeta.
* Add a --drop-prefixes flag to swift-account-info,
swift-container-info, and swift-object-info. This makes the
output between the three more consistent.
* statsd error messages correspond to 5xx responses only. This
makes monitoring more useful because actual errors (5xx) will
not be hidden by common user requests (4xx). Previously, some 4xx
responses would be included in timing information in the statsd
error messages.
* Truncate error logs to prevent log handler from running out of buffer.
* Updated requirements.txt to match global exclusions and formatting.
* tempauth user names now support unicode characters.
* Various other minor bug fixes and improvements.
swift (2.17.1, queens stable backports)
* Fix SLO delete for accounts with non-ASCII names.
* Fixed an issue in COPY where concurrent requests may have copied the
wrong data.
* Fixed a bug in how Swift uses eventlet that was exposed under high
concurrency.
swift (2.17.0, OpenStack Queens)
* Added symlink objects support.
Symlink objects reference one other object. They are created by
creating an empty object with an X-Symlink-Target header. The value of
the header is of the format <container>/<object>, and the target does
not need to exist at the time of symlink creation. Cross-account
symlinks can be created by including the
X-Symlink-Target-Account header.
GET and HEAD requests to a symlink will operate on the
referenced object and require appropriate permission in the
target container. DELETE and PUT requests will operate on the
symlink object itself. POST requests are not forwarded to the
referenced object. POST requests sent to a symlink will result
in a 307 Temporary Redirect response.
* Added support for inline data segments in SLO manifests.
Upgrade impact: during a rolling upgrade, an updated proxy server
may write a manifest that an out-of-date proxy server will not be
able to read. This will resolve itself once the upgrade completes
on all nodes.
* The tempurl digest algorithm is now configurable, and Swift added
support for both SHA-256 and SHA-512. Supported tempurl digests
are exposed to clients in `/info`. Additionally, tempurl signatures
can now be base64 encoded.
* Object expiry improvements
- Disallow X-Delete-At header values equal to the X-Timestamp header.
- X-Delete-At computation now uses X-Timestamp instead of
system time. This prevents clock skew causing inconsistent
expiry data.
- Deleting an expiring object will now cause less work in the system.
The number of async pending files written has been reduced for all
objects and greatly reduced for erasure-coded objects. This
dramatically reduces the burden on container servers.
- Stopped logging tracebacks when receiving an unexpected response.
- Allow the expirer to gracefully move past updating stale work items.
* When the object auditor examines an object, it will now add any
missing metadata checksums.
* `swift-ring-builder` improvements
- Save the ring when dispersion improves, even if balance
doesn't improve.
- Improved the granularity of the ring dispersion metric so that
small improvements after a rebalance can show changes in the
dispersion number. Dispersion in existing and new rings can be
recalculated using the new '--recalculate' option to
`swift-ring-builder`.
- Display more info on empty rings.
* Fixed rare socket leak on range requests to erasure-coded objects.
* The number of container updates on object PUTs (ie to update listings)
has been recomputed to be far more efficient while maintaining
durability guarantees. Specifically, object PUTs to erasure-coded
policies will now normally result in far fewer container updates.
* Moved Zuul v3 tox jobs into the Swift code repo.
* Changed where liberasurecode-devel for CentOS 7 is referenced and
installed as a dependency.
* Added container/object listing with prefix to InternalClient.
* Added '--swift-versions' to `swift-recon` CLI to compare installed
versions in the cluster.
* Stop logging tracebacks in the `object-replicator` when it runs
out of handoff locations.
* Send ETag header in 206 Partial Content responses to SLO reads.
* Now `swift-recon-cron` works with conf.d configs.
* Improved `object-updater` stats logging. It now tells you all of
its stats (successes, failures, quarantines due to bad pickles,
unlinks, and errors), and it tells you incremental progress every
five minutes. The logging at the end of a pass remains and has
been expanded to also include all stats.
* If a proxy server is configured to autocreate accounts and the
account create fails, it will now return a server error (500)
instead of Not Found (404).
* Fractional replicas are no longer allowed for erasure code policies.
* Various other minor bug fixes and improvements.
swift (2.16.0)
* Add checksum to object extended attributes.
* Let clients request heartbeats during SLO PUTs by including
the query parameter `heartbeat=on`.
With heartbeating turned on, the proxy will start its response
immediately with 202 Accepted then send a single whitespace
character periodically until the request completes. At that
point, a final summary chunk will be sent which includes a
"Response Status" key indicating success or failure and (if
successful) an "Etag" key indicating the Etag of the resulting
SLO.
* Added support for retrieving the encryption root secret from an
external key management system. In practice, this is currently limited
to Barbican.
* Move listing formatting out to a new proxy middleware named
`listing_formats`. `listing_formats` should be just right of the
first proxy-logging middleware, and left of most other
middlewares. If it is not already present, it will be
automatically inserted for you.
Note: if you have a custom middleware that makes account or
container listings, it will only receive listings in JSON format.
* Log deprecation warning for `allow_versions` in the container
server config. Configure the `versioned_writes` middleware in
the proxy server instead. This option will be ignored in a
future release.
* Replaced `replication_one_per_device` by custom count defined by
`replication_concurrency_per_device`. The original config value
is deprecated, but continues to function for now. If both values
are defined, the old `replication_one_per_device` is ignored.
* Fixed a rare issue where multiple backend timeouts could result
in bad data being returned to the client.
* Cleaned up logged tracebacks when talking to memcached servers.
* Account and container replication stats logs now include
`remote_merges`, the number of times a whole database was sent
to another node.
* Respond 400 Bad Request when Accept headers fail to parse
instead of returning 406 Not Acceptable.
* The `domain_remap` middleware now supports the
`mangle_client_paths` option. Its default "false" value changes
`domain_remap` parsing to stop stripping the `path_root` value
from URL paths. If users depend on this path mangling, operators
should set `mangle_client_paths` to "True" before upgrading.
* Remove `swift-temp-url` script. The functionality has been in
swiftclient for a long time and this script has been deprecated
since 2.10.0.
* Removed all `post_as_copy` related code and configs. The option
has been deprecated since 2.13.0.
* Fixed XML responses (eg on bulk extractions and SLO upload
failures) to be more correct. The enclosing "delete" tag was
removed where it doesn't make sense and replaced with "extract"
or "upload" depending on the context.
* Static Large Object (SLO) manifest may now (again) have zero-byte
last segments.
* Fixed an issue where background consistency daemon child
processes would deadlock waiting on the same file descriptor.
* Removed a race condition where a POST to an SLO could modify the
X-Static-Large-Object metadata.
* Accept a trade off of dispersion for balance in the ring builder
that will result in getting to balanced rings much more quickly
in some cases.
* Fixed using `swift-ring-builder set_weight` with more than one
device.
* When requesting objects, return 404 if a tombstone is found and
is newer than any data found. Previous behavior was to return
stale data.
* Various other minor bug fixes and improvements.
swift (2.15.2, pike stable backports)
* Fixed a cache invalidation issue related to GET and PUT requests to
containers that would occasionally cause object PUTs to a container to
404 after the container had been successfully created.
* Removed a race condition where a POST to an SLO could modify the
X-Static-Large-Object metadata.
* Fixed rare socket leak on range requests to erasure-coded objects.
* Fix SLO delete for accounts with non-ASCII names.
* Fixed an issue in COPY where concurrent requests may have copied the
wrong data.
* Fixed time skew when using X-Delete-After.
* Send ETag header in 206 Partial Content responses to SLO reads.
swift (2.15.1, OpenStack Pike)
* Fixed a bug introduced in 2.15.0 where the object reconstructor
would exit with a traceback if no EC policy was configured.
* Fixed deadlock when logging from a tpool thread.
The object server runs certain IO-intensive methods outside the
main pthread for performance. Previously, if one of those methods
tried to log, this can cause a crash that eventually leads to an
object server with hundreds or thousands of greenthreads, all
deadlocked. The fix is to use a mutex that works across different
greenlets and different pthreads.
* The object reconstructor can now rebuild an EC fragment for an
expired object.
* Various other minor bug fixes and improvements.
swift (2.15.0)
* Add Composite Ring Functionality
A composite ring comprises two or more component rings that are
combined to form a single ring with a replica count equal to the
sum of the component rings. The component rings are built
independently, using distinct devices in distinct regions, which
means that the dispersion of replicas between the components can
be guaranteed.
Composite rings can be used for explicit replica placement and
"replicated EC" for global erasure codes policies.
Composite rings support 'cooperative' rebalance which means that
during rebalance all component rings will be consulted before a
partition is moved in any component ring. This avoids the same
partition being simultaneously moved in multiple components.
We do not yet have CLI tools for creating composite rings, but
the functionality has been enabled in the ring modules to
support this advanced functionality. CLI tools will be delivered
in a subsequent release.
For further information see the docs at
<https://docs.openstack.org/swift/latest/overview_ring.html#module-swift.common.ring.composite_builder>
* The EC reconstructor process has been dramatically improved by
adding support for multiple concurrent workers. Multiple
processes are required to get high concurrency, and this change
results in much faster rebalance times on servers with many
drives.
Currently the default is still only one process, and no workers.
Set `reconstructor_workers` in the `[object-reconstructor]`
section to some whole number <= the number of devices on a node
to get that many reconstructor workers.
* Add support to increase object ring partition power transparently
to end users and with no cluster downtime. Increasing the ring
partition power allows for incremental adjustment to the upper bound
of the cluster size. Please review the full docs at
<https://docs.openstack.org/swift/latest/ring_partpower.html>.
* Added support for per-policy proxy config options. This allows
per-policy affinity options to be set for use with duplicated EC
policies and composite rings. Certain options found in per-policy
conf sections will override their equivalents that may be set
in the [app:proxy-server] section. Currently the options handled that
way are sorting_method, read_affinity, write_affinity,
write_affinity_node_count, and write_affinity_handoff_delete_count.
* Enabled versioned writes on Dynamic Large Objects (DLOs).
* Write-affinity aware object deletion
Previously, when deleting objects in multi-region swift
deployment with write affinity configured, users always get 404
when deleting object before it's replicated to appropriate nodes.
Now Swift will use `write_affinity_handoff_delete_count` to
define how many local handoff nodes should swift send request to
get more candidates for the final response. The default value
"auto" means Swift will calculate the number automatically based
on the number of replicas and current cluster topology.
* Require that known-bad EC schemes be deprecated
Erasure-coded storage policies using isa_l_rs_vand and nparity
>= 5 must be configured as deprecated, preventing any new
containers from being created with such a policy. This
configuration is known to harm data durability. Any data in such
policies should be migrated to a new policy. See
https://bugs.launchpad.net/swift/+bug/1639691 for more
information
* Optimize the Erasure Code reconstructor protocol to reduce IO
load on servers.
* Fixed a bug where SSYNC would fail to replicate unexpired object.
* Fixed a bug in domain_remap when obj starts/ends with slash.
* Fixed a socket leak in copy middleware when a large object was copied.
* Fixed a few areas where the `swiftdir` option was not respected.
* `swift-recon` now respects storage policy aliases.
* cname_lookup middleware now accepts a `nameservers` config
variable that, if defined, will be used for DNS lookups instead of
the system default.
* Make mount_check option usable in containerized environments by
adding a check for an ".ismount" file at the root directory of
a device.
* Remove deprecated `vm_test_mode` option.
* The object and container server config option `slowdown` has been
deprecated in favor of the new `objects_per_second` and
`containers_per_second` options.
* The output of devices from `swift-ring-builder` has been reordered
by region, zone, ip, and device.
* Imported docs content from openstack-manuals project.
* Various other minor bug fixes and improvements.
swift (2.14.0)
* Fixed error where a container drive error resulted in double space
usage on rest drives. When drive with container or account database
is unmounted, the bug would create handoff replicas on all remaining
drives, increasing the drive space used and filling the cluster.
* Fixed UnicodeDecodeError in the object reconstructor that would
prevent objects with non-ascii names from being reconstructed and
caused the reconstructor process to hang.
* EC Fragment Duplication - Foundational Global EC Cluster Support.
* Fixed encoding issue in ssync where a mix of ascii and non-ascii
metadata values would cause an error.
* `name_check` and `cname_lookup` keys have been added to `/info`.
* Add Vary: headers for CORS responses.
* Always set Swift processes to use UTC.
* Prevent logged traceback in object-server on client disconnect for
chunked transfers to replicated policies.
* Removed per-device reconstruction stats. Now that the reconstructor
is shuffling parts before going through them, those stats no longer
make sense.
* Log correct status code for conditional requests.
* Drop support for auth-server from common/manager.py and `swift-init`.
* Include received fragment index in reconstructor log warnings.
* Fixed a race condition in updating hashes.pkl where a partition
suffix invalidation may have been skipped.
* `domain_remap` now accepts a list of domains in "storage_domain".
* Do not follow CNAME when host is in storage_domain.
* Enable cluster-wide CORS Expose-Headers setting via
"cors_expose_headers".
* Cache all answers from nameservers in cname_lookup.
* Log the correct request type of a subrequest downstream of copy.
* Various other minor bug fixes and improvements.
swift (2.13.0, OpenStack Ocata)
* Improvements in key parts of the consistency engine
- Improved performance by eliminating an unneeded directory
structure hash.
- Optimized the common case for hashing filesystem trees, thus
eliminating a lot of extraneous disk I/O.
- Updated the `hashes.pkl` file format to include timestamp information
for race detection. Also simplified hashing logic to prevent race
conditions and optimize for the common case.
- The erasure code reconstructor will now shuffle work jobs across all
disks instead of going disk-by-disk. This eliminates single-disk I/O
contention and allows continued scaling as concurrency is increased.
- Erasure code reconstruction handles moving data from handoff nodes
better. Instead of moving the data to another handoff, it waits
until it can be moved to a primary node.
Upgrade Impact: If you upgrade and roll back, you must delete all
`hashes.pkl` files.
* If using erasure coding with ISA-L in rs_vand mode and 5 or more parity
fragments, Swift will emit a warning. This is a configuration that is
known to harm data durability. In a future release, this warning will be
upgraded to an error unless the policy is marked as deprecated. All data
in an erasure code storage policy using isa_l_rs_vand with 5 or more
parity should be migrated as soon as possible. Please see
https://bugs.launchpad.net/swift/+bug/1639691 for more information.
* The erasure code reconstructor `handoffs_first` option has been
deprecated in favor of `handoffs_only`. `handoffs_only` is far more
useful, and just like `handoffs_first` mode in the replicator, it gives
the operator the option of forcing the consistency engine to focus
solely on revert (handoff) jobs, thus improving the speed of
rebalances. The `handoffs_only` behavior is somewhat consistent with
the replicator's `handoffs_first` option (any error on any handoff in
the replicator will make it essentially handoff only forever) but the
`handoff_only` option does what you want and is named correctly in the
reconstructor.
* The default for `object_post_as_copy` has been changed to False. The
option is now deprecated and will be removed in a future release. If
your cluster is still running with post-as-copy enabled, please update
it to use the "fast-post" method. Future versions of Swift will not
support post-as-copy, and future features will not be supported under
post-as-copy. ("Fast-post" is where `object_post_as_copy` is false).
* Temporary URLs now support one common form of ISO 8601 timestamps in
addition to Unix seconds-since-epoch timestamps. The ISO 8601 format
accepted is '%Y-%m-%dT%H:%M:%SZ'. This makes TempURLs more
user-friendly to produce and consume.
* Listing containers in accounts with json or xml now includes a
`last_modified` time. This does not change any on-disk data, but simply
exposes the value to offer consistency with the object listings on
containers.
* Fixed a bug where the ring builder would not allow removal of a device
when min_part_seconds_left was greater than zero.
* PUT subrequests generated from a client-side COPY will now properly log
the SSC (server-side copy) Swift source field. See
https://docs.openstack.org/swift/latest/logs.html#swift-source for
more information.
* Fixed a bug where an SLO download with a range request may have resulted
in a 5xx series response.
* SLO manifest PUT requests can now be properly validated by sending an
ETag header of the md5 sum of the concatenated md5 sums of the
referenced segments.
* Fixed the stats calculation in the erasure code reconstructor.
* Rings with min_part_hours set to zero will now only move one partition
replica per rebalance, thus matching behavior when min_part_hours is
greater than zero.
* I/O priority is now supported on AArch64 architecture.
* Various other minor bug fixes and improvements.
swift (2.12.0)
* Ring files now include byteorder information about the endian of
the machine used to generate the file, and the values are
appropriately byteswapped if deserialized on a machine with a
different endianness.
Newly created ring files will be byteorder agnostic, but
previously generated ring files will still fail on different
endian architectures. Regenerating older ring files will cause
them to become byteorder agnostic. The regeneration of the ring
files will not cause any new data movement. Newer ring files
will still be usable by older versions of Swift (on machines
with the same endianness--this maintains existing behavior).
* All 416 responses will now include a Content-Range header with
an unsatisfied-range value. This allows the caller to know the
valid range request value for an object.
* TempURLs now support a validation against a common prefix. A
prefix-based signature grants access to all objects which share the
same prefix. This avoids the creation of a large amount of signatures,
when a whole container or pseudofolder is shared.
* Correctly handle deleted files with if-none-match requests.
* Correctly send 412 Precondition Failed if a user sends an
invalid copy destination. Previously Swift would send a 500
Internal Server Error.
* In SLO manifests, the `etag` and `size_bytes` keys are now fully
optional and not required. Previously, the keys needed to exist
but the values were optional. The only required key is `path`.
* Fixed a rare infinite loop in `swift-ring-builder` while placing parts.
* Ensure update of the container by object-updater, removing a rare
possibility that objects would never be added to a container listing.
* Fixed non-deterministic suffix updates in hashes.pkl where a partition
may be updated much less often than expected.
* Fixed regression in consolidate_hashes that occurred when a new
file was stored to new suffix to a non-empty partition. This bug
was introduced in 2.7.0 and could cause an increase in rsync
replication stats during and after upgrade, due to inconsistent
hashing of partition suffixes.
* Account and container databases will now be quarantined if the
database schema has been corrupted.
* Removed "in-process-" from func env tox name to work with
upstream CI.
* Respect server type for --md5 check in swift-recon.
* Remove empty db hash and suffix directories if a db gets quarantined.
* Various other minor bug fixes and improvements.
swift (2.11.0)
* We have made significant improvements and changes to the erasure
code implementation.
- Instead of using a separate .durable file to indicate the
durable status of an EC fragment archive, we rename the .data
to include a durable marker in the filename. This saves one
inode for every EC .data file. Existing .durable files will not
be removed, and they will continue to work just fine.
Note that after writing EC data with Swift 2.11.0 or later, that
data will not be accessible to earlier versions of Swift.
- Closed a bug where ssync may have written bad fragment data in
some circumstances. A check was added to ensure the correct number
of bytes is written for a fragment before finalizing the write.
Also, erasure coded fragment metadata will now be validated on read
requests and, if bad data is found, the fragment will be quarantined.
- The improvements to EC reads made in Swift 2.10.0 have also been
applied to the reconstructor. This allows fragments to be rebuilt
in more circumstances, resulting in faster recovery from failures.
- WARNING: If you are using the ISA-L library for erasure codes,
please upgrade to liberasurecode 1.3.1 (or later) as soon as
possible. If you are using isa_l_rs_vand with more than 4 parity,
please read https://bugs.launchpad.net/swift/+bug/1639691 and take
necessary action.
- Updated the PyECLib dependency to 1.3.1.
* Added a configurable URL base to staticweb.
* Support multi-range GETs for static large objects.
* TempURLs using the "inline" parameter can now also set the
"filename" parameter. Both are used in the Content-Disposition
response header.
* Mirror X-Trans-Id to X-Openstack-Request-Id.
* SLO will now concurrently HEAD segments, resulting in much faster
manifest validation and object creation. By default, two HEAD requests
will be done at a time, but this can be changed by the operator via
the new `concurrency` setting in the "[filter:slo]" section of
the proxy server config.
* Suppressed the KeyError message when auditor finds an expired object.
* Daemons using InternalClient can now be properly killed with SIGTERM.
* Added a "user" option to the drive-audit config file. Its value is
used to set the owner of the drive-audit recon cache.
* Throttle update_auditor_status calls so it updates no more than once
per minute.
* Suppress unexpected-file warnings for rsync temp files.
* Various other minor bug fixes and improvements.
swift (2.10.0, OpenStack Newton)
* Object versioning now supports a "history" mode in addition to
the older "stack" mode. The difference is in how DELETE requests
are handled. For full details, please read
https://docs.openstack.org/swift/latest/overview_object_versioning.html.
* New config variables to change the schedule priority and I/O
scheduling class. Servers and daemons now understand
`nice_priority`, `ionice_class`, and `ionice_priority` to
schedule their relative importance. Please read
https://docs.openstack.org/swift/latest/admin_guide.html
for full config details.
* On newer kernels (3.15+ when using xfs), Swift will use the O_TMPFILE
flag when opening a file instead of creating a temporary file
and renaming it on commit. This makes the data path simpler and
allows the filesystem to more efficiently optimize the files on
disk, resulting in better performance.
* Erasure code GET performance has been significantly
improved in clusters that are not completely healthy.
* Significant improvements to the api-ref doc available at
https://docs.openstack.org/api-ref/object-store/.
* A PUT or POST to a container will now update the container's
Last-Modified time, and that value will be included in a
GET/HEAD response.
* Include object sysmeta in POST responses. Sysmeta is still
stripped from the response before being sent to the client, but
this allows middleware to make use of the information.
* Fixed a bug where a container listing delimiter wouldn't work
with encryption.
* Fixed a bug where some headers weren't being copied correctly
in a COPY request.
* Container sync can now copy SLOs more efficiently by allowing
the manifest to be synced before all of the referenced segments.
This fixes a bug where container sync would not copy SLO manifests.
* Fixed a bug where some tombstone files might never be reclaimed.
* Update dnspython dependency to 1.14, removing the need to have
separate dnspython dependencies for Py2 and Py3.
* Deprecate swift-temp-url and call python-swiftclient's
implementation instead. This adds python-swiftclient as an
optional dependency of Swift.
* Moved other-requirements.txt to bindep.txt. bindep.txt lists
non-python dependencies of Swift.
* Various other minor bug fixes and improvements.
swift (2.9.0)
* Swift now supports at-rest encryption. This feature encrypts all
object data and user-set object metadata as it is sent to the cluster.
This feature is designed to prevent information leaks if a hard drive
leaves the cluster. The encryption is transparent to the end-user.
At-rest encryption in Swift is enabled on the proxy server by
adding two middlewares to the pipeline. The `keymaster` middleware
is responsible for managing the encryption keys and the `encryption`
middleware does the actual encryption and decryption.
Existing clusters will continue to work without enabling
encryption. Although enabling this feature on existing clusters
is supported, best practice is to enable this feature on new
clusters when the cluster is created.
For more information on the details of the at-rest encryption
feature, please see the docs at
https://docs.openstack.org/swift/latest/overview_encryption.html.
* `swift-recon` can now be called with more than one server type.
* Fixed a bug where non-ascii names could cause an error in logging
and cause a 5xx response to the client.
* The install guide and API reference have been moved into Swift's
source code repository.
* Various other minor bug fixes and improvements.
swift (2.8.0)
* Allow concurrent bulk deletes for server-side deletes of static
large objects. Previously this would be single-threaded and each
DELETE executed serially. The new `delete_concurrency` value
(default value is 2) in the `[filter:slo]` and `[filter:bulk]`
sections of the proxy server config controls the concurrency
used to perform the DELETE requests for referenced segments. The
default value is recommended, but setting the value to 1
restores previous behavior.
* Refactor server-side copy as middleware
The COPY verb is now implemented in the `copy` middleware instead
of in the proxy server code. If not explicitly added, the server
side copy middleware is auto-inserted to the left of `dlo`, `slo`
and `versioned_writes` middlewares in the proxy server pipeline.
As a result, dlo and slo `copy_hooks` are no longer required. SLO
manifests are now validated when copied so when copying a
manifest to another account the referenced segments must be
readable in that account for the manifest copy to succeed
(previously this validation was not made, meaning the manifest
was copied but could be unusable if the segments were not
readable).
With this change, there should be no change in functionality or
existing behavior.
* `fallocate_reserve` can now be a percentage (a value ending in "%"),
and the default has been adjusted to "1%".
* Now properly require account/container metadata be valid UTF-8
* TempURL responses now include an `Expires` header with the
expiration time embedded in the URL.
* Non-Python dependencies are now listed in other-requirements.txt.
* `swift-ring-builder` now supports a `--yes` option to assume a
yes response to all questions. This is useful for scripts.
* Write requests to a replicated storage policy with an even number
of replicas now have a quorum size of half the replica count
instead of half-plus-one.
* Container sync now logs per-container stat information so operators
can track progress. This is logged at INFO level.
* `swift-dispersion-*` now allows region to be specified when there
are multiple Swift regions served by the same Keystone instance
* Fix infinite recursion during logging when syslog is down.
* Fixed a bug where a backend failure during a read could result in
a missing byte in the response body.
* Stop `staticweb` revealing container existence to unauth'd requests.
* Reclaim isolated .meta files if they are older than the `reclaim_age`.
* Make `rsync` ignore its own temporary files instead of spreading
them around the cluster, wasting space.
* The object auditor now ignores files in the devices directory when
auditing objects.
* The deprecated `threads_per_disk` setting has been removed. Deployers
are encouraged to use `servers_per_port` instead.
* Fixed an issue where a single-replica configuration for account or
container DBs could result in the DB being inadvertently deleted if
it was placed on a handoff node.
* `disable_fallocate` now also correctly disables `fallocate_reserve`.
* Fixed a bug where the account-reaper did not delete all containers
in a reaped account.
* Correctly handle delimiter queries where results start with the
delimiter and no prefix is given.
* Changed the recommended ports for Swift services from ports
6000-6002 to unused ports 6200-6202 so they do not conflict with
X-Windows or other services. Since these config values must be
explicitly set in the config file, this doesn't impact existing
deployments.
* Fixed an instance where REPLICATE requests would not use
`replication_ip`.
* Various other minor bug fixes and improvements.
swift (2.7.0, OpenStack Mitaka)
* Bump PyECLib requirement to >= 1.2.0
* Update container on fast-POST
"Fast-POST" is the mode where `object_post_as_copy` is set to
`False` in the proxy server config. This mode now allows for
fast, efficient updates of metadata without needing to fully
recopy the contents of the object. While the default still is
`object_post_as_copy` as True, the plan is to change the default
to False and then deprecate post-as-copy functionality in later
releases. Fast-POST now supports container-sync functionality.
* Add concurrent reads option to proxy.
This change adds 2 new parameters to enable and control concurrent
GETs in Swift, these are `concurrent_gets` and `concurrency_timeout`.
`concurrent_gets` allows you to turn on or off concurrent
GETs; when on, it will set the GET/HEAD concurrency to the
replica count. And in the case of EC HEADs it will set it to
ndata. The proxy will then serve only the first valid source to
respond. This applies to all account, container, and replicated
object GETs and HEADs. For EC only HEAD requests are affected.
The default for `concurrent_gets` is off.
`concurrency_timeout` is related to `concurrent_gets` and is
the amount of time to wait before firing the next thread. A
value of 0 will fire at the same time (fully concurrent), but
setting another value will stagger the firing allowing you the
ability to give a node a short chance to respond before firing
the next. This value is a float and should be somewhere between
0 and `node_timeout`. The default is `conn_timeout`, meaning by
default it will stagger the firing.
* Added an operational procedures guide to the docs. It can be
found at https://docs.openstack.org/swift/latest/ops_runbook/index.html and
includes information on detecting and handling day-to-day
operational issues in a Swift cluster.
* Make `handoffs_first` a more useful mode for the object replicator.
The `handoffs_first` replication mode is used during periods of
problematic cluster behavior (e.g. full disks) when replication
needs to quickly drain partitions from a handoff node and move
them to a primary node.
Previously, `handoffs_first` would sort that handoff work before
"normal" replication jobs, but the normal replication work could
take quite some time and result in handoffs not being drained
quickly enough.
In order to focus on getting handoff partitions off the node
`handoffs_first` mode will now abort the current replication
sweep before attempting any primary suffix syncing if any of the
handoff partitions were not removed for any reason - and start
over with replication of handoffs jobs as the highest priority.
Note that `handoffs_first` being enabled will emit a warning on
start up, even if no handoff jobs fail, because of the negative
impact it can have during normal operations by dog-piling on a
node that was temporarily unavailable.
* By default, inbound `X-Timestamp` headers are now disallowed
(except when in an authorized container-sync request). This
header is useful for allowing data migration from other storage
systems to Swift and keeping the original timestamp of the data.
If you have this migration use case (or any other requirement on
allowing the clients to set an object's timestamp), set the
`shunt_inbound_x_timestamp` config variable to False in the
gatekeeper middleware config section of the proxy server config.
* Requesting a SLO manifest file with the query parameters
"?multipart-manifest=get&format=raw" will return the contents of
the manifest in the format as was originally sent by the client.
The "format=raw" is new.
* Static web page listings can now be rendered with a custom
label. By default listings are rendered with a label of:
"Listing of /v1/<account>/<container>/<path>". This change adds
a new custom metadata key/value pair
`X-Container-Meta-Web-Listings-Label: My Label` that when set,
will cause the following: "Listing of My Label/<path>" to be
rendered instead.
* Previously, static large objects (SLOs) had a minimum segment
size (default to 1MiB). This limit has been removed, but small
segments will be ratelimited. The config parameter
`rate_limit_under_size` controls the definition of "small"
segments (1MiB by default), and `rate_limit_segments_per_sec`
controls how many segments per second can be served (default is 1).
With the default values, the effective behavior is identical to the
previous behavior when serving SLOs.
* Container sync has been improved to perform a HEAD on the remote
side of the sync for each object being synced. If the object
exists on the remote side, container-sync will no longer
transfer the object, thus significantly lowering the network
requirements to use the feature.
* The object auditor will now clean up any old, stale rsync temp
files that it finds. These rsync temp files are left if the
rsync process fails without completing a full transfer of an
object. Since these files can be large, the temp files may end
up filling a disk. The new auditor functionality will reap these
rsync temp files if they are old. The new object-auditor config
variable `rsync_tempfile_timeout` is the number of seconds old a
tempfile must be before it is reaped. By default, this variable
is set to "auto" or the rsync_timeout plus 900 seconds (falling
back to a value of 1 day).
* The Erasure Code reconstruction process has been made more
efficient by not syncing data files when only the durable commit
file is missing.
* Fixed a bug where 304 and 416 response may not have the right
Etag and Accept-Ranges headers when the object is stored in an
Erasure Coded policy.
* Versioned writes now correctly stores the date of previous versions
using GMT instead of local time.
* The deprecated Keystone middleware option is_admin has been removed.
* Fixed log format in object auditor.
* The zero-byte mode (ZBF) of the object auditor will now properly
observe the `--once` option.
* Swift keeps track, internally, of "dirty" parts of the partition
keyspace with a "hashes.pkl" file. Operations on this file no
longer require a read-modify-write cycle and use a new
"hashes.invalid" file to track dirty partitions. This change
will improve end-user performance for PUT and DELETE operations.
* The object replicator's succeeded and failed counts are now logged.
* `swift-recon` can now query hosts by storage policy.
* The log_statsd_host value can now be an IPv6 address or a hostname
which only resolves to an IPv6 address.
* Erasure coded fragments now properly call fallocate to reserve disk
space before being written.
* Various other minor bug fixes and improvements.
swift (2.6.0)
* Dependency changes
- Updated minimum version of eventlet to 0.17.4 to support IPv6.
- Updated the minimum version of PyECLib to 1.0.7.
* The ring rebalancing algorithm was updated to better handle edge cases
and to give better (more balanced) rings in the general case. New rings
will have better initial placement, capacity adjustments will move less
data for better balance, and existing rings that were imbalanced should
start to become better balanced as they go through rebalance cycles.
* Added container and account reverse listings.
A GET request to an account or container resource with a "reverse=true"
query parameter will return the listing in reverse order. When
iterating over pages of reverse listings, the relative order of marker
and end_marker are swapped.
* Storage policies now support having more than one name.
This allows operators to fix a typo without breaking existing clients,
or, alternatively, have "short names" for policies. This is implemented
with the "aliases" config key in the storage policy config in
swift.conf. The aliases value is a list of names that the storage
policy may also be identified by. The storage policy "name" is used to
report the policy to users (eg in container headers). The aliases have
the same naming restrictions as the policy's primary name.
* The object auditor learned the "interval" config value to control the
time between each audit pass.
* `swift-recon --all` now includes the config checksum check.
* `swift-init` learned the --kill-after-timeout option to force a service
to quit (SIGKILL) after a designated time.
* `swift-recon` now correctly shows timestamps in UTC instead of local
time.
* Fixed bug where `swift-ring-builder` couldn't select device id 0.
* Documented the previously undocumented
`swift-ring-builder pretend_min_part_hours_passed` command.
* The "node_timeout" config value now accepts decimal values.
* `swift-ring-builder` now properly removes devices with zero weight.
* `swift-init` return codes are updated via "--strict" and "--non-strict"
options. Please see the usage string for more information.
* `swift-ring-builder` now reports the min_part_hours lockout time
remaining
* Container sync has been improved to more quickly find and iterate over
the containers to be synced. This reduced server load and lowers the
time required to see data propagate between two clusters. Please see
https://docs.openstack.org/swift/latest/overview_container_sync.html for more details
about the new on-disk structure for tracking synchronized containers.
* A container POST will now update that container's put-timestamp value.
* TempURL header restrictions are now exposed in /info.
* Error messages on static large object manifest responses have been
greatly improved.
* Closed a bug where an unfinished read of a large object would leak a
socket file descriptor and a small amount of memory. (CVE-2016-0738)
* Fixed an issue where a zero-byte object PUT with an incorrect Etag
would return a 503.
* Fixed an error when a static large object manifest references the same
object more than once.
* Improved performance of finding handoff nodes if a zone is empty.
* Fixed duplication of headers in Access-Control-Expose-Headers on CORS
requests.
* Fixed handling of IPv6 connections to memcache pools.
* Continued work towards python 3 compatibility.
* Various other minor bug fixes and improvements.
swift (2.5.0, OpenStack Liberty)
* Added the ability to specify ranges for Static Large Object (SLO)
segments.
* Replicator configs now support an "rsync_module" value to allow
for per-device rsync modules. This setting gives operators the
ability to fine-tune replication traffic in a Swift cluster and
isolate replication disk IO to a particular device. Please see
the docs and sample config files for more information and
examples.
* Significant work has gone in to testing, fixing, and validating
Swift's erasure code support at different scales.
* Swift now emits StatsD metrics on a per-policy basis.
* Fixed an issue with Keystone integration where a COPY request to a
service account may have succeeded even if a service token was not
included in the request.
* Ring validation now warns if a placement partition gets assigned to the
same device multiple times. This happens when devices in the ring are
unbalanced (e.g. two servers where one server has significantly more
available capacity).
* Various other minor bug fixes and improvements.
swift (2.4.0)
* Dependency changes
- Added six requirement. This is part of an ongoing effort to add
support for Python 3.
- Dropped support for Python 2.6.
* Config changes
- Recent versions of Python restrict the number of headers allowed in a
request to 100. This number may be too low for custom middleware. The
new "extra_header_count" config value in swift.conf can be used to
increase the number of headers allowed.
- Renamed "run_pause" setting to "interval" (current configs with
run_pause still work). Future versions of Swift may remove the
run_pause setting.
* Versioned writes middleware
The versioned writes feature has been refactored and reimplemented as
middleware. You should explicitly add the versioned_writes middleware to
your proxy pipeline, but do not remove or disable the existing container
server config setting ("allow_versions"), if it is currently enabled.
The existing container server config setting enables existing
containers to continue being versioned. Please see
https://docs.openstack.org/swift/latest/middleware.html#how-to-enable-object-versioning-in-a-swift-cluster
for further upgrade notes.
* Allow 1+ object-servers-per-disk deployment
Enabled by a new > 0 integer config value, "servers_per_port" in the
[DEFAULT] config section for object-server and/or replication server
configs. The setting's integer value determines how many different
object-server workers handle requests for any single unique local port
in the ring. In this mode, the parent swift-object-server process
continues to run as the original user (i.e. root if low-port binding
is required), binds to all ports as defined in the ring, and forks off
the specified number of workers per listen socket. The child, per-port
servers drop privileges and behave pretty much how object-server workers
always have, except that because the ring has unique ports per disk, the
object-servers will only be handling requests for a single disk. The
parent process detects dead servers and restarts them (with the correct
listen socket), starts missing servers when an updated ring file is
found with a device on the server with a new port, and kills extraneous
servers when their port is found to no longer be in the ring. The ring
files are stat'ed at most every "ring_check_interval" seconds, as
configured in the object-server config (same default of 15s).
In testing, this deployment configuration (with a value of 3) lowers
request latency, improves requests per second, and isolates slow disk
IO as compared to the existing "workers" setting. To use this, each
device must be added to the ring using a different port.
* Do container listing updates in another (green)thread
The object server has learned the "container_update_timeout" setting
(with a default of 1 second). This value is the number of seconds that
the object server will wait for the container server to update the
listing before returning the status of the object PUT operation.
Previously, the object server would wait up to 3 seconds for the
container server response. The new behavior dramatically lowers object
PUT latency when container servers in the cluster are busy (e.g. when
the container is very large). Setting the value too low may result in a
client PUT'ing an object and not being able to immediately find it in
listings. Setting it too high will increase latency for clients when
container servers are busy.
* TempURL fixes (closes CVE-2015-5223)
Do not allow PUT tempurls to create pointers to other data.
Specifically, disallow the creation of DLO object manifests via a PUT
tempurl. This prevents discoverability attacks which can use any PUT
tempurl to probe for private data by creating a DLO object manifest and
then using the PUT tempurl to head the object.
* Ring changes
- Partition placement no longer uses the port number to place
partitions. This improves dispersion in small clusters running one
object server per drive, and it does not affect dispersion in
clusters running one object server per server.
- Added ring-builder-analyzer tool to more easily test and analyze a
series of ring management operations.
- Stop moving partitions unnecessarily when overload is on.
* Significant improvements and bug fixes have been made to erasure code
support. This feature is suitable for beta testing, but it is not yet
ready for broad production usage.
* Bulk upload now treats user xattrs on files in the given archive as
object metadata on the resulting created objects.
* Emit warning log in object replicator if "handoffs_first" or
"handoff_delete" is set.
* Enable object replicator's failure count in swift-recon.
* Added storage policy support to dispersion tools.
* Support keystone v3 domains in swift-dispersion.
* Added domain_remap information to the /info endpoint.
* Added support for a "default_reseller_prefix" in domain_remap
middleware config.
* Allow SLO PUTs to forgo per-segment integrity checks. Previously, each
segment referenced in the manifest also needed the correct etag and
bytes setting. These fields now allow the "null" value to skip those
particular checks on the given segment.
* Allow rsync to use compression via a "rsync_compress" config. If set to
true, compression is only enabled for an rsync to a device in a
different region. In some cases, this can speed up cross-region
replication data transfer.
* Added time synchronization check in swift-recon (the --time option).
* The account reaper now runs faster on large accounts.
* Various other minor bug fixes and improvements.
swift (2.3.0, OpenStack Kilo)
* Erasure Code support (beta)
Swift now supports an erasure-code (EC) storage policy type. This allows
deployers to achieve very high durability with less raw capacity as used
in replicated storage. However, EC requires more CPU and network
resources, so it is not good for every use case. EC is great for storing
large, infrequently accessed data in a single region.
Swift's implementation of erasure codes is meant to be transparent to
end users. There is no API difference between replicated storage and
EC storage.
To support erasure codes, Swift now depends on PyECLib and
liberasurecode. liberasurecode is a pluggable library that allows for
the actual EC algorithm to be implemented in a library of your choosing.
As a beta release, EC support is nearly fully feature complete, but it
is lacking support for some features (like multi-range reads) and has
not had a full performance characterization. This feature relies on
ssync for durability. Deployers are urged to do extensive testing and
not deploy production data using an erasure code storage policy.
Full docs are at https://docs.openstack.org/swift/latest/overview_erasure_code.html
* Add support for container TempURL Keys.
* Make more memcache options configurable. connection_timeout,
pool_timeout, tries, and io_timeout are all now configurable.
* Swift now supports composite tokens. This allows another service to
act on behalf of a user, but only with that user's consent.
See https://docs.openstack.org/swift/latest/overview_auth.html for more details.
* Multi-region replication was improved. When replicating data to a
different region, only one replica will be pushed per replication
cycle. This gives the remote region a chance to replicate the data
locally instead of pushing more data over the inter-region network.
* Internal requests from the ratelimit middleware now properly log a
swift_source. See https://docs.openstack.org/swift/latest/logs.html for details.
* Improved storage policy support for quarantine stats in swift-recon.
* The proxy log line now includes the request's storage policy index.
* Ring checker has been added to swift-recon to validate if rings are
built correctly. As part of this feature, storage servers have learned
the OPTIONS verb.
* Add support of x-remove- headers for container-sync.
* Rings now support hostnames instead of just IP addresses.
* Swift now enforces that the API version on a request is valid. Valid
versions are configured via the valid_api_versions setting in swift.conf
* Various other minor bug fixes and improvements.
swift (2.2.2)
* Data placement changes
This release has several major changes to data placement in Swift in
order to better handle different deployment patterns. First, with an
unbalance-able ring, less partitions will move if the movement doesn't
result in any better dispersion across failure domains. Also, empty
(partition weight of zero) devices will no longer keep partitions after
rebalancing when there is an unbalance-able ring.
Second, the notion of "overload" has been added to Swift's rings. This
allows devices to take some extra partitions (more than would normally
be allowed by the device weight) so that smaller and unbalanced clusters
will have less data movement between servers, zones, or regions if there
is a failure in the cluster.
Finally, rings have a new metric called "dispersion". This is the
percentage of partitions in the ring that have too many replicas in a
particular failure domain. For example, if you have three servers in a
cluster but two replicas for a partition get placed onto the same
server, that partition will count towards the dispersion metric. A
lower value is better, and the value can be used to find the proper
value for "overload".
The overload and dispersion metrics have been exposed in the
swift-ring-build CLI tools.
See https://docs.openstack.org/swift/latest/overview_ring.html
for more info on how data placement works now.
* Improve replication of large out-of-sync, out-of-date containers.
* Added console logging to swift-drive-audit with a new log_to_console
config option (default False).
* Optimize replication when a device and/or partition is specified.
* Fix dynamic large object manifests getting versioned. This was not
intended and did not work. Now it is properly prevented.
* Fix the GET's response code when there is a missing segment in a
large object manifest.
* Change black/white listing in ratelimit middleware to use sysmeta.
Instead of using the config option, operators can set
"X-Account-Sysmeta-Global-Write-Ratelimit: WHITELIST" or
"X-Account-Sysmeta-Global-Write-Ratelimit: BLACKLIST" on an account to
whitelist or blacklist it for ratelimiting. Note: the existing
config options continue to work.
* Use TCP_NODELAY on outgoing connections.
* Improve object-replicator startup time.
* Implement OPTIONS verb for storage nodes.
* Various other minor bug fixes and improvements.
swift (2.2.1)
* Swift now rejects object names with Unicode surrogates.
* Return 403 (instead of 413) on unauthorized upload when over account
quota.
* Fix a rare condition when a rebalance could cause swift-ring-builder
to crash. This would only happen on old ring files when "rebalance"
was the first command run.
* Storage node error limits now survive a ring reload.
* Speed up reading and writing xattrs for object metadata by using larger
xattr value sizes. The change is moving from 254 byte values to 64KiB
values. There is no migration issue with this.
* Deleted containers beyond the reclaim age are now properly reclaimed.
* Full Simplified Chinese translation (zh_CN locale) for errors and logs.
* Container quota is now properly enforced during cross-account COPY.
* ssync replication now properly uses the configured replication_ip.
* Fixed issue were ssync did not replicate custom object headers.
* swift-drive-audit now has the 'unmount_failed_device' config option
(default to True) that controls if the process will unmount failed
drives or not.
* swift-drive-audit will now dump drive error rates to a recon file.
The file location is controlled by the 'recon_cache_path' config value
and it includes each drive and its associated number of errors.
* When a filesystem does't support xattr, the object server now returns
a 507 Insufficient Storage error to the proxy server.
* Clean up empty account and container partitions directories if they
are empty. This keeps the system healthy and prevents a large number
of empty directories from slowing down the replication process.
* Show the sum of every policy's amount of async pendings in swift-recon.
* Various other minor bug fixes and improvements.
swift (2.2.0, OpenStack Juno)
* Added support for Keystone v3 auth.
Keystone v3 introduced the concept of "domains" and user names
are no longer unique across domains. Swift's Keystone integration
now requires that ACLs be set on IDs, which are unique across
domains, and further restricts setting new ACLs to only use IDs.
Please see https://docs.openstack.org/swift/latest/overview_auth.html for
more information on configuring Swift and Keystone together.
* Swift now supports server-side account-to-account copy. Server-
side copy in Swift requires the X-Copy-From header (on a PUT)
or the Destination header (on a COPY). To initiate an account-to-
account copy, the existing header value remains the same, but the
X-Copy-From-Account header (on a PUT) or the Destination-Account
(on a COPY) are used to indicate the proper account.
* Limit partition movement when adding a new placement tier.
When adding a new placement tier (server, zone, or region), Swift
previously attempted to move all placement partitions, regardless
of the space available on the new tier, to ensure the best possible
durability. Unfortunately, this could result in too many partitions
being moved all at once to a new tier. Swift's ring-builder now
ensures that only the correct number of placement partitions are
rebalanced, and thus makes adding capacity to the cluster more
efficient.
* Per storage policy container counts are now reported in an
account response headers.
* Swift will now reject, with a 4xx series response, GET requests
with more than 50 ranges, more than 3 overlapping ranges, or more
than 8 non-increasing ranges.
* The bind_port config setting is now required to be explicitly set.
* The object server can now use splice() for a zero-copy GET
response. This feature is enabled with the "splice" config variable
in the object server config and defaults to off. Also, this feature
only works on recent Linux kernels (AF_ALG sockets must be
supported). A zero-copy GET response can significantly reduce CPU
requirements for object servers.
* Added "--no-overlap" option to swift-dispersion populate so that
multiple runs of the tool can add coverage without overlapping
existing monitored partitions.
* swift-recon now supports filtering by region.
* Various other minor bug fixes and improvements.
swift (2.1.0)
* swift-ring-builder placement was improved to allow gradual addition
of new regions without causing a massive migration of data to the new
region. The change was to prefer device weight first, then look at
failure domains.
* Logging updates
- Eliminated "Handoff requested (N)" log spam.
- Added process pid to the end of storage node log lines.
- Container auditor now logs a warning if the devices path contains a
non-directory.
- Object daemons now send a user-agent string with their full name.
* 412 and 416 responses are no longer tracked as errors in the StatsD
messages from the backend servers.
* Parallel object auditor
The object auditor can now be controlled with a "concurrency" config
value that allows multiple auditor processes to run at once. Using
multiple parallel auditor processes can speed up the overall auditor
cycle time.
* The object updater will now concurrently update each necessary node
in a new greenthread.
* TempURL updates
- The default allowed methods have changed to also allow POST and
DELETE. The new default list is "GET HEAD PUT POST DELETE".
- TempURLs for POST now also allow HEAD, matching existing GET and PUT
functionality.
- Added filename*= support to TempURL Content-Disposition response
header.
* X-Delete-At/After can now be used with the FormPost middleware.
* Make swift-form-signature output a sample form.
* Add v2 API to list endpoints middleware
The new API adds better support for storage policies and changes the
response from a list of backend urls to a dictionary with the keys
"endpoints" and "headers". The endpoints key contains a list of the
backend urls, and the headers key is a dictionary of headers to send
along with the backend request.
* Added allow_account_management and account_autocreate values to /info
responses.
* Enable object system metadata on PUTs (Note: POST support is ongoing).
* Various other minor bug fixes and improvements.
swift (2.0.0)
* Storage policies
Storage policies allow deployers to configure multiple object rings
and expose them to end users on a per-container basis. Deployers
can create policies based on hardware performance, regions, or other
criteria and independently choose different replication factors on
them. A policy is set on a Swift container at container creation
time and cannot be changed.
Full docs are at https://docs.openstack.org/swift/latest/overview_policies.html
* Add profiling middleware in Swift
The profile middleware provides a tool to profile Swift
code on the fly and collects statistical data for performance
analysis. A native simple Web UI is also provided to help
query and visualize the data.
* Add --quoted option to swift-temp-url
* swift-recon now supports checking the md5sum of swift.conf, which
helps deployers verify configurations are consistent across a cluster.
* Users can now set the transaction id suffix by passing in
a value in the X-Trans-Id-Extra header.
* New log_max_line_length option caps the maximum length of a log line.
* Support If-[Un]Modified-Since for object HEAD
* Added missing constraints and ratelimit parameters to /info
* Add ability to remove subsections from /info
* Unify logging for account, container, and object server processes
to provide a consistent message format. This change reorders the
fields logged for the account server.
* Add targeted config loading to swift-init. This allows an easier
and more explicit way to tell swift-init to run specific server
process configurations.
* Properly quote www-authenticate (CVE-2014-3497)
* Fix logging issue when services stop on py26.
* Change the default logged length of the auth token to 16.
* Explicitly set permissions on generated ring files to 0644
* Fix file uploads larger than 2GiB in the formpost feature
* Fixed issue where large objects would fail to download if the
auth token expired partway through the download
* Various other minor bug fixes and improvements
swift (1.13.1, OpenStack Icehouse)
* Change the behavior of CORS responses to better match the spec
A new proxy config variable (strict_cors_mode, default to True)
has been added. Setting it to False keeps the old behavior. For
an overview of old versus new behavior, please see
https://review.opendev.org/#/c/69419/
* Invert the responsibility of the two instances of proxy-logging in
the proxy pipeline
The first proxy_logging middleware instance to receive a request
in the pipeline marks that request as handling it. So now, the
left most proxy_logging middleware handles logging for all
client requests, and the right most proxy_logging middleware
handles all other requests initiated from within the pipeline to
its left. This fixes logging related to large object
requests not properly recording bandwidth.
* Added swift-container-info and swift-account-info tools
* Allow specification of object devices for audit
* Dynamic large object COPY requests with ?multipart-manifest=get
now work as expected
* When a client is downloading a large object and one of the segment
reads gets bad data, Swift will now immediately abort the request.
* Fix ring-builder crash when a ring partition was assigned to a
deleted device, zero-weighted device, and normal device
* Make probetests work with conf.d configs
* Various other minor bug fixes and improvements.
swift (1.13.0)
* Account-level ACLs and ACL format v2
Accounts now have a new privileged header to represent ACLs or
any other form of account-level access control. The value of
the header is a JSON dictionary string to be interpreted by the
auth system. A reference implementation is given in TempAuth.
Please see the full docs at
https://docs.openstack.org/swift/latest/overview_auth.html
* Added a WSGI environment flag to stop swob from always using
absolute location. This is useful if middleware needs to use
out-of-spec Location headers in a response.
* Container sync proxies now support simple load balancing
* Config option to lower the timeout for recoverable object GETs
* Add a way to ratelimit all writes to an account
* Allow multiple storage_domain values in cname_lookup middleware
* Moved all DLO functionality into middleware
The proxy will automatically insert the dlo middleware at an
appropriate place in the pipeline the same way it does with the
gatekeeper middleware. Clusters will still support DLOs after upgrade
even with an old config file that doesn't mention dlo at all.
* Remove python-swiftclient dependency
* Add secondary groups to process user during privilege escalation
* When logging request headers, it is now possible to specify
specifically which headers should be logged
* Added log_requests config parameter to account and container servers
to match the parameter in the object server. This allows a deployer
to turn off log messages for these processes.
* Ensure swift.source is set for DLO/SLO requests
* Fixed an issue where overwriting segments in a dynamic manifest
could cause issues on pipelined requests.
* Properly handle COPY verb in container quota middleware
* Improved StaticWeb 404 error message on web-listings and index
* Various other minor bug fixes and improvements.
swift (1.12.0)
* Several important pieces of information have been added to /info:
- Configured constraints are included and allow a client to discover
the limits on names and object sizes that the cluster supports.
- The supported tempurl methods are now included.
- Static large object constraints are now included.
* The Last-Modified header value returned will now be the object's
timestamp rounded up to the next second. This allows subsequent
requests with If-[un]modified-Since to use the Last-Modified
value as expected.
* Non-integer values for if-delete-at headers will now properly
report a 400 error instead of a 503.
* Fix object versioning with non-ASCII container names.
* Bulk delete with POST now works properly.
* Generic means for persisting system metadata
Swift now supports system-level metadata on accounts and
containers. System metadata provides a means to store internal
custom metadata with associated Swift resources in a safe and
secure fashion without actually having to plumb custom metadata
through the core swift servers. The new gatekeeper middleware
prevents this system metadata from leaking into the request or
being set by a client.
* catch_errors and gatekeeper middleware are now forced into the proxy
pipeline if not explicitly referenced.
* New container sync configuration option, separating the end user
from knowing the required end point and adding more secure
signed requests. See
https://docs.openstack.org/swift/latest/overview_container_sync.html
for full information.
* bulk middleware now can be configured to retry deleting containers.
* The default yield_frequency used to keep client connections alive
during slow bulk requests was reduced from 60 seconds to 10 seconds.
While this is a change to a default, it should not affect deployments
and there is no migration process needed.
* Swift processes will attempt to set RLIMIT_NPROC to 8192.
* Server processes will now exit with a non-zero error code on config
errors.
* Warn if read_affinity is configured but not enabled.
* Fix checkmount error parsing in swift-recon.
* Log at warn level when an object is quarantined.
* Fixed CVE-2014-0006 to avoid a potential timing attack with tempurl.
* Various other minor bug fixes and improvements.
swift (1.11.0)
* Added discoverable capabilities
A Swift proxy server now by default (although it can be turned off)
will respond to requests to /info. The response to these requests
include information about the cluster and can be used by clients to
determine which features are supported in the cluster.
* Object replication ssync (an rsync alternative)
A Swift storage node can now be configured to use Swift primitives
for replication transport instead of rsync. This is an experimental
feature that is not yet considered production ready.
* If a source times out on an object server read, try another one
of them with a modified range.
* The proxy now responds to many types of requests as soon as it
has a quorum. This can help speed up responses (without
changing the results), especially when one node is acting up.
There is a post_quorum_timeout config value that can tune how
long to wait for requests to finish after a quorum has been
established.
* Add accurate timestamps in proxy log lines for the start and
end of a request. These are added as new fields on the end of
the existing log lines, and therefore should not break
existing, well-behaved log processors.
* Add an "inline" query parameter to tempurl
By default, temporary URLs add a "Content-Disposition" header
that forces many clients to download the object. Now, temporary
URLs support an optional "inline" query parameter that will
force a "Content-Disposition: inline" header to be added to the
response, overriding the default.
* Use TCP_NODELAY for created sockets. This can dramatically
lower latency for small object workloads.
* DiskFile API, with reference implementation
The DiskFile abstraction for talking to data on disk has been
refactored to allow alternate implementations to be developed.
Included in the codebase is an in-memory reference
implementation. For full documentation, please see the developer
documentation. The DiskFile API is still a work in progress and
is not yet finalized.
* Removal of swift-bench
The included benchmarking tool swift-bench has been extracted
from the codebase and is now in its own repository at
https://github.com/openstack/swift-bench. New swift-bench
binaries and packages may be found on PyPI at
https://pypi.org/project/swift-bench
* Bulk delete now also supports the POST verb, in addition to DELETE
* Added functionality to the swift-ring-builder to support
limited recreation of ring builder files from the ring file itself.
* HEAD on account now returns 410 if account was deleted and
not yet reaped. The old behavior was to return a 404.
* Fixed a bug introduced since the 1.10.0 release that
prevented expired objects from being removed from the system.
This resulted in orphaned expired objects taking up space on
the system but inaccessible to the API. This regression and
fix are only important if you have deployed code since the
1.10.0 release. For a full discussion, including a script that
can be used to clean up orphaned objects, see
https://bugs.launchpad.net/swift/+bug/1257330
* Tie socket write buffer size to server chunk size parameter. This
pairs the underlying network buffer size with the size of data
that Swift attempts to read from the connection, thereby
improving efficiency and throughput on connections.
* Fix 500 from account-quota middleware. If a user had set
X-Account-Meta-Quota-Bytes to something non-integer prior to
the installation of the account-quota middleware, then the
quota check would choke on it. Now a non-integer value is
treated as "no quota".
* Quarantine objects with busted metadata. Before, if you
encountered an object with corrupt or missing xattrs, the
object server would return a 500 on GET, and wouldn't quarantine
anything. Now the object server returns a 404 for that GET and
the corrupted file is quarantined, thus giving replication a
chance to fix it.
* Fix quarantine and error counts in audit logs
* Report transaction ID in failure exception logs
* Make pbr a build-time only dependency
* Worked around a bug in eventlet 0.9.16 where the size of the
memcache connection pools would grow unbounded.
* Tempurl keys are now properly stored as utf8
* Fixed an issue where concurrent PUT requests to accounts or
containers may result in errors due to locked databases.
* Handle copy requests in account and container quota middleware
* Now ensure that a WWW-Authenticate header is on all 401 responses
* Various other bug fixes and improvements
swift (1.10.0, OpenStack Havana)
* Added support for pooling memcache connections
* Added support to replicating handoff partitions first in object
replication. Can also configure how many remote nodes a storage node
must talk to before removing a local handoff partition.
* Fixed bug where memcache entries would not expire
* Much faster calculation for choosing handoff nodes
* Added container listing ratelimiting
* Fixed issue where the proxy would continue to read from a storage
server even after a client had disconnected
* Added support for headers that are only visible to the owner of a Swift
account
* Fixed ranged GET with If-None-Match
* Fixed an issue where rings may not be balanced after initial creation
* Fixed internationalization support
* Return the correct etag for a static large object on the PUT response
* Allow users to extract archives to containers with ACLs set
* Fix support for range requests against static large objects
* Now logs x-copy-from header in a useful place
* Reverted back to old XML output of account and container listings to
ensure older clients do not break
* Account quotas now appropriately handle copy requests
* Fix issue with UTF-8 handling in versioned writes
* Various other bug fixes and improvements, including support for running
Swift under Pypy and continuing work to support storage policies
swift (1.9.1)
* Disallow PUT, POST, and DELETE requests from creating older tombstone
files, preventing the possibility of filling up the disk and removing
unnecessary container updates.
* Set default wsgi workers to cpu_count
Change the default value of wsgi workers from 1 to auto. The new
default value for workers in the proxy, container, account & object
wsgi servers will spawn as many workers per process as you have cpu
cores. This will not be ideal for some configurations, but it's much
more likely to produce a successful out of the box deployment.
* Added reveal_sensitive_prefix config setting to filter the auth token
logged by the proxy server.
* Ensure Keystone's reseller prefix ends with an underscore. Previously
this was a recommendation--now it is enforced.
* Added log_file_pattern config to swift-drive-audit for drive errors
* Add support for telling Swift to detect a content type on a request.
* Additional object stats are now logged in the object auditor
* Moved the DiskFile interface into its own module
* Ensure the SQLite cursors are closed when creating functions
* Better support for valid Accept headers
* In Keystone, don't allow users to delete their own account
* Return a UTC timezone designator in container listings
* Ensure that users can't remove their account quotas
* Allow floating point value for dispersion coverage
* Fix incorrect error page handling in staticweb
* Add utf-8 charset to multipart-manifest=get response.
* Allow dispersion tools to use keystone server with insecure certificate
* Ensure that files are always closed in tests
* Use OpenStack's "Hacking" guidelines for code formatting
* Various other minor bug fixes and improvements
swift (1.9.0)
* Global clusters support
The "region" concept introduced in Swift 1.8.0 has been augmented with
support for using a separate replication network and configuring read
and write affinity. These features combine to offer support for a single
Swift cluster spanning wide geographic area.
* Disk performance
The object server now can be configured to use threadpools to increase
performance and smooth out latency throughout the system. Also, many
disk operations were reordered to increase reliability and improve
performance.
* Added config file conf.d support
Allow Swift daemons and servers to optionally accept a directory as the
configuration parameter. This allows different parts of the config file
to be managed separately, eg each middleware could use a separate file
for its particular config settings.
* Allow two TempURL keys per account
By adding a second key, a user can safely rotate keys and prevent URLs
already in use from becoming invalid. TempURL middlware has also been
updated to allow a configuable set of allowed methods and to prevent a
bugrelated to content-disposition names.
* Added crossdomain.xml middleware. See
https://docs.openstack.org/swift/latest/crossdomain.html for details
* Added rsync bandwidth limit setting for object replicator
* Transaction ID updated to include the time and an optional suffix
* Added x-remove-versions-location header to disable versioned writes
* Improvements to support for Keystone ACLs
* Added parallelism to object expirer daemon
* Added support for ring hash prefix in addition to the existing suffix
* Allow all headers requested for CORS
* Stop getting useless bytes on manifest Range requests
* Improved container-sync resiliency
* Added example Apache config files. See
https://docs.openstack.org/swift/latest/apache_deployment_guide.html
for more info
* If an account is marked as deleted but hasn't been reaped and is still
on disk, responses will include an "X-Account-Status" header
* Fix 503 on account/container HEAD with invalid format
* Added extra safety on account-level DELETE when using bulk deletes
* Made colons quote-safe in logs (mainly for IPv6)
* Fixed bug with bulk delete max items
* Fixed static large object manifest range requests
* Prevent static large objects from containing other static large objects
* Fixed issue with use of delimiter in container queries where some
objects would not be listed
* Various other minor bug fixes and improvements
swift (1.8.0, OpenStack Grizzly)
* Make rings' replica count adjustable
* Added a region tier to the ring above zones
* Added timing-based sorting of object servers on read requests
* Added support for auto-extract archive uploads
* Added support for bulk delete requests
* Added support for large objects with static manifests
* Added list_endpoints middleware to provide an API for determining where
the ring places data
* proxy-logging middleware can now handle logging for other middleware
proxy-logging should be used twice in the proxy pipeline. The first
handles middleware logs for requests that never made it all the way
to the server. The last handles requests that do make it to the server.
This is a change that may require an update to your proxy server
config file or custom middleware that you may be using. See the full
docs at https://docs.openstack.org/swift/latest/misc.html.
* Changed the default sample rate for a few high-traffic requests.
Added log_statsd_sample_rate_factor to globally tune the StatsD
sample rate. This tunable can be used to reduce StatsD traffic
proportionally for all metrics and is intended to replace
log_statsd_default_sample_rate, which is left alone for
backward-compatibility, should anyone be using it.
* Added swift_hash_path_prefix option to swift.conf
New deployments are advised to set this value to a random secret
to protect against hash collisions
* Added user-managed container quotas
* Added support for account-level quotas managed by an auth reseller
* Added --run-dir option to swift-init
* Added more options to swift-bench
* Added support for CORS "actual requests"
* Added fallocate_reserve option to protect against full drives
* Allow ring rebalance to take a seed
* Ring serialization will now produce the same gzip file (Py2.7)
* Added support to swift-drive-audit for handling rotated logs
* Added first-byte latency timings for GET requests
* Added per disk PUT timing monitoring support
* Added speed limit options for DB auditor
* Force log entries to be one line
* Ensure that fsync is used and not just fdatasync
* Improved handoff node selection
* Deprecated keystone is_admin feature
* Fix large objects with unicode in the segment names
* Update Swift's MemcacheRing to provide API compatibility with
standard Python memcache libraries
* Various other minor bug fixes and improvements
swift (1.7.6)
* Better tempauth storage URL guessing
* Added --top option to swift-recon -d
* Allow optional, temporary healthcheck failure
* keystoneauth middleware now supports cross-tenant ACLs
* Add dispersion report flags to limit reports
* Add config option to turn eventlet debug on/off
* Added override option for swift-init's KILL_WAIT
* Added oldest and most recent replication pass to swift-recon
* Fixed 500 error response when GETing a many-segment manifest
* Memcached keys now use a delta timeout when possible
* Refactor DiskFile to hide temp file names and exts
* Remove IP-based container-sync ACLs from auth middlewares
* Fixed bug in deleting memcached account info data
* Fixed lazy-listing of object manifest segments
* Fixed bug where a ? in the object name caused an error
* Swift now returns 406 if it can't satisfy Accept
* Fix infinite recursion bug in object replicator
* Swift will now reject names with NULL characters
* Fixed object-auditor logging to use a minimum of unix sockets
* Various other minor bug fixes and improvements
swift (1.7.5)
* Support OPTIONS verb, including CORS preflight requests
* Added support for custom log handlers
* Range support is extended to support GET requests with multiple ranges.
Multi-range GETs are not yet supported against large-object manifests.
* Cluster constraints are now settable by config
* Replicators can now run against specific devices or partitions
* swift-bench now supports running on multiple cores and multiple servers
* Added partition option to swift-get-nodes
* Allow underscores in account and user in tempauth via base64 encodings
* New option to the dispersion report to output the missing partitions
* Changed storage server StatsD metrics to report timings instead of
counts for errors. See the admin guide for the updated metric names.
* Removed a dependency on WebOb and replaced it with an internal module
* Fixed config parsing in swift-bench -x
* Fixed sample_rate in StatsD logging
* Track unlinks of async_pendings with StatsD
* Remove double GET on range requests
* Allow unsetting of X-Container-Sync-To and ACL headers
* DB reclamation now removes empty suffix directories
* Fix non-standard 100-continue behavior
* Allow object-expirer to delete the last copy of a versioned object
* Only set TCP_KEEPIDLE on systems where it is supported
* Fix stdin flush and fdatasync issues on BSD platforms
* Allow object-expirer to delete the last version of an object
* Various other minor bug fixes and improvements
swift (1.7.4, OpenStack Folsom)
* Fix issue where early client disconnects may have caused a memory leak
swift (1.7.2)
* Fix issue where memcache serialization was not properly loading
the config value
swift (1.7.0)
* Use custom encoding for ring data instead of pickle
Serialize RingData in a versioned, custom format which is a combination
of a JSON-encoded header and .tostring() dumps of the
replica2part2dev_id arrays. This format deserializes hundreds of times
faster than rings serialized with Python 2.7's pickle (a significant
performance regression for ring loading between Python 2.6 and Python
2.7). Fixes bug 1031954.
The new implementation is backward-compatible; if a ring
does not begin with a new-style magic string, it is assumed to be an
old-style pickle-dumped ring and is handled as before. So new Swift
code can read old rings, but old Swift code will not be able to read
newly-serialized rings.
* Do not use pickle for serialization in memcache, but JSON
To avoid issues on upgrades (unability to read pickled values, and cache
poisoning for old servers not understanding JSON), we add a
memcache_serialization_support configuration option, with the following
values:
0 = older, insecure pickle serialization
1 = json serialization but pickles can still be read (still insecure)
2 = json serialization only (secure and the default)
To avoid an instant full cache flush, existing installations should
upgrade with 0, then set to 1 and reload, then after some time (24
hours) set to 2 and reload. Support for 0 and 1 will be removed in
future versions.
* Update proxy-server StatsD logging. This is a significant change to the
existing StatsD intigration. Docs for this feature can be found in
doc/source/admin_guide.rst.
* Improved swift-bench to allow random object sizes and better usability
* Updated probe tests
* Replicator removal metrics are now generated on a per-device basis
* Made object replicator locking more optimistic
* Split proxy-server code into separate modules
* Fixed bug where swift-recon would not report all unmounted drives
* Fixed issue where a LockTimeout may have caused a file descriptor to
not be closed properly
* Fixed a bug where an error may have caused the proxy to stop returning
data to a client
* Fixed bug where expirer would get confused by odd deletion times
* Fixed a bug where auto-creating accounts would return an error if they
were recreated after being deleted
* Fix when rate_limit_after_segment kicks in
* fallocate() failures properly return HTTPInsufficientStorage from
object-server before reading from wsgi.input, allowing the proxy
server to quickly error_limit that node
* Fixed error with large object manifests and x-newest headers on GET
* Various other minor bug fixes and improvements
swift (1.6.0)
* Removed bin/swift and swift/common/client.py from the swift repo. These
tools are now managed in the python-swiftclient project. The
python-swiftclient project is a second deliverable of the openstack
swift project.
* Moved swift_auth (openstack keystone) middleware from keystone project
into swift project
* Made dispersion report work with any replica count other than 3. This
substantially affects the JSON output of the dispersion report, and any
tools written to consume this output will need to be updated.
* Added Solaris (Illumos) compatibility
* Added -a option to swift-get-nodes to show all handoffs
* Add UDP protocol support for logger
* Added config options for rate limiting of large object downloads.
* Added config option `log_handoffs` (defaults to True) to proxy server
to log and update statsd with information about when a handoff node is
used. This is helpful to track the health of the cluster.
* swift-bench can now use auth 2.0
* Support forbidding substrings based on a regexp in name_filter
middleware
* Hardened internal server processes so only authorized methods can be
called.
* Made ranged requests on large objects work correctly when size of
manifest file is not 0 byte
* Added option to dispersion report to print 404s to stdout
* Fix object replication on older rsync versions when using ipv4
* Fixed bug with container reclaim/report race
* Make object server's caching more configurable.
* Check disk failure before syncing for each partition
* Allow special characters to be referenced by manifest objects
* Validate devices and partitions to avoid directory traversals
* Support WebOb 1.2
* Ensure that accessing the ring devs reloads the ring if necessary.
Specifically, this allows replication to work when it has been started
with an empty ring.
* Various other minor bug fixes and improvements
swift (1.5.0)
* New option to toggle SQLite database preallocation with account
and container servers.
IMPORTANT:
The default for database preallocation is now off when before
it was always on. This will affect performance on clusters that
use standard drives with shared account, container, object
servers. Such deployments will need to update their
configurations to turn database preallocation back on (see
account-server.conf-sample and container-server.conf.sample
files).
If you are using dedicated account and container servers with
SSDs, you should defragment your file systems after upgrade and
should notice dramatically less disk usage.
* swift3 middleware removed and moved to http://github.com/fujita/swift3.
This will require a config change in the proxy server and adds a new
dependency for deployers using this middleware.
* Moved proxy server logging to middleware. This requires a config change
in the proxy server.
* Added object versioning feature. (See docs for full description)
* Add statsd logging throughout the system (beta, some event names may
change)
* Expanded swift-recon middleware support
* The ring builder now supports as-unique-as-possible partition
placement, unified balancing methods, and can work on more than one
device at a time.
* Numerous bug fixes to StaticWeb (previously unusable at scale).
* Bug fixes to all middleware to allow passthrough requests under various
conditions and to share pre-authed request code (which previously had
differing behaviors and interaction bugs).
* Bug fix to object expirer that could cause infinite looping.
* Added optional delay to account reaping.
* Async-pending write optimization.
* Dispersion tools now support multiple auth versions
* Updated man pages
* Proxy server can now deny requests to particular hostnames
* Updated docs for domain remap middleware
* Updated docs for cname lookup middleware
* Made swift CLI binary easier to wrap
* Proxy will now also return X-Timestamp header
* Added associated projects doc as a place to track ecosystem projects
* end_marker made consistent across both object and container listings
* Various other minor bug fixes and improvements
swift (1.4.8, OpenStack Essex)
* Added optional max_containers_per_account restriction
* Added alternate metadata header removal method
* Added optional name_check middleware filter
* Added support for venv-based test runs with tox
* StaticWeb behavior change with X-Web-Mode: true and
non-StaticWeb-enabled containers (immediately 404s instead of passing
the request on down the WSGI pipeline).
* Fixed typo in swift-dispersion-report JSON output.
* Swift-Recon-related fix to create temporary files on the same disk as
their final destinations.
* Updated return codes in swift3 middleware
* Fixed swift3 middleware to allow Content-Range header in response
* Updated swift.common.client and swift CLI tool with auth 2.0 changes
* Swift CLI tool now supports common openstack auth args
* Body of HTTP responses now included in error messages of swift CLI tool
* Refactored some ring building functions for clarity and simplicity
swift (1.4.7)
* Improvements to account and container replication.
* Fix for account servers allowing .pending to exist before .db.
* Fixed possible key-guessing exploit in formpost.
* Fixed bug in ring builder when removing a large percentage of devices.
* Swift CLI tool now supports openstack-standard CLI flags.
* New JSON output option for swift-dispersion-report.
* Removed old stats tools.
* Other bug fixes and documentation updates.
swift (1.4.6)
* TempURL and FormPost middleware added
* Added memcache.conf option
* Dropped eval-based json parser fallback
* Properly lose all groups when dropping privileges
* Fix permissions when creating files
* Fixed bug regarding negative Content-Length in requests
* Consistent formatting on Last-Modified response header
* Added timeout option to swift-recon
* Allow arguments to be passed to nosetest
* Removed tools/rfc.sh
* Other minor bug fixes
swift (1.4.5)
* New swift-orphans and swift-oldies command line tools to detect
orphaned Swift processes and long running processes.
* Command line tool "swift" now supports marker queries.
* StaticWeb middleware improved to save an extra request when
possible.
* Updated swift-init to support swift-object-expirer.
* Fixed object replicator timeout handling [bug 814263].
* Fixed accept header 503 vs. 400 [bug 891247].
* More exception handling for auditors.
* Doc updates for PPA [bug 905608].
* Doc updates to explain replication more clearly [bug 906976].
* Updated SAIO instructions to no longer mention ~/swift/trunk.
* Fixed docstrings in the ring code.
* PEP8 Updates.
swift (1.4.4)
* Fixes to prevent socket hoarding (memory leak)
* Add sockstat info to recon.
* Fixed leak from SegmentedIterable.
* Fixed bufferedhttp to deref socks and fps.
* Add support for OS Auth API version 2.
* Make Eventlet's WSGI server log differently.
* Updated TimeoutError and except Exception refs.
* Fixed time-sensitive tests.
* Fixed object manifest etags.
* Fixes for swift-recon disk usage distribution graph.
* Adding new manpages for configuration files.
* Change bzr to swift in getting_started doc.
* Fixes the HTTPConflict import.
* Expiring Objects Support.
* Fixing bug with x-trans-id.
* Requote the source when doing a COPY.
* Add documentation for Swift Recon.
* Make drive audit regexes detect 4-letter drives.
* Adding what acc/cont/obj into the ratelimit error messages.
* Query only specific zone via swift-recon.
swift (1.4.3, OpenStack Diablo)
* Additional quarantine catching code.
* Added client_ip to all proxy log lines not otherwise containing it.
* Content-Type is now application/xml for "GET services/bucket" swift3
middleware requests.
* Alpha release of the Swift Recon Experiment
* Fix last modified date for swift3 middleware.
* Fix to clear account/container metadata on account/container deletion.
* Fix for corner case regarding X-Newest.
* Fix for object auditor running out of file descriptors.
* Fix to return all proper headers for manifest objects.
* Fix to the swift tool to strip any leading slashes on file names when
uploading.
swift (1.4.2)
* Removed stats/logging code from Swift [now in separate slogging project].
* Container Synchronization Feature - First Edition
* Fix swift3 authentication bug about the Date and X-Amz-Date handling.
* Changing ratelimiting so that it only limits PUTs/DELETEs.
* Object POSTs are implemented as COPYs now by default (you can revert to
previous implementation with conf object_post_as_copy = false)
* You can specify X-Newest: true on GETs and HEADs to indicate you want
Swift to query all backend copies and return the newest version
retrieved.
* Object COPY requests now always copy the newest object they can find.
* Account and container GETs and HEADs now shuffle the nodes they use to
balance load.
* Fixed the infinite charset: utf-8 bug
* This fixes the bug that drop_buffer_cache() doesn't work on systems where
off_t isn't 64 bits.
swift (1.4.1)
* st renamed to swift
* swauth was separated froms swift. It is now its own project and can be
found at https://github.com/gholt/swauth.
* tempauth middleware added as an extremely limited auth system for dev
work.
* Account and container listings now properly labeled UTF-8 (previously the
label was "utf8").
* Accounts are auto-created if an auth token is valid when the
account_autocreate proxy config parameter is set to true.
swift (1.4.0)
* swift-bench now cleans up containers it creates.
* WSGI servers now load WSGI filters and applications after forking for
better plugin support.
* swauth-cleanup-tokens now handles 404s on token containers and tokens
better.
* Proxy logs the remote IP address as the client IP in the absence of
X-Forwarded-For and X-Cluster-Client-IP headers instead of - like it did
before.
* Swift3 WSGI middleware added support for param-signed URLs.
* swauth- scripts now exit with proper exit codes.
* Fixed a bug where allowed_headers weren't honored for HEAD requests.
* Double quarantining of corrupted sqlite3 databases now works.
* Fix for Object replicator breaking when running object replicator with no
objects on the server.
* Added the Accept-Ranges header to GET and HEAD requests.
* When a single object has multiple async pending updates on a single
device, only latest async pending is now sent.
* Fixed issue of Swift3 WSGI middleware not working correctly with '/' in
object names.
* Renamed swift-stats-* to swift-dispersion-* to avoid confusion with log
stats stuff.
* Added X-Trans-Id transaction id header to every response.
* Fixed a Python 2.7 compatibility problem.
* Now using bracketed notation for ip literals in rsync calls, so
compressed ipv6 literals work.
* Added a container stats collector and refactoring some of the stats code.
* Changed subdir nodes in XML formatted object listings to align with
object nodes. Now: <subdir name="foo"><name>foo</name></subdir> Before:
<subdir name="foo" />.
* Fixed bug in Swauth to support for multiple swauth instances.
* swift-ring-builder: Added list_parts command which shows common
partitions for a given list of devices.
* Object auditor now shows better statistics updates in the logs.
* Stats uploaders now allow overrides for source_filename_pattern and
new_log_cutoff values.
----
Changelog entries for previous versions are incomplete
swift (1.3.0, OpenStack Cactus)
swift (1.2.0, OpenStack Bexar)
swift (1.1.0, OpenStack Austin)
swift (1.0.0, Initial Release)
View Git Blame Copy Permalink