openstack
/
swift
Code Issues Proposed changes

Add Storage Policy Documentation 

Add overview and example information for using Storage Policies.

DocImpact
Implements: blueprint storage-policies
Change-Id: I6f11f7a1bdaa6f3defb3baa56a820050e5f727f1
This commit is contained in:
Paul Luse 2014-04-07 14:22:27 -07:00 committed by Clay Gerrard
parent c1dc2fa624
commit e52e8bc917
13 changed files with 923 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
doc/saio/bin/remakerings
View File

@ -10,6 +10,12 @@ swift-ring-builder object.builder add r1z2-127.0.0.1:6020/sdb2 1
swift-ring-builder object.builder add r1z3-127.0.0.1:6030/sdb3 1
swift-ring-builder object.builder add r1z4-127.0.0.1:6040/sdb4 1
swift-ring-builder object.builder rebalance
swift-ring-builder object-1.builder create 10 2 1
swift-ring-builder object-1.builder add r1z1-127.0.0.1:6010/sdb1 1
swift-ring-builder object-1.builder add r1z2-127.0.0.1:6020/sdb2 1
swift-ring-builder object-1.builder add r1z3-127.0.0.1:6030/sdb3 1
swift-ring-builder object-1.builder add r1z4-127.0.0.1:6040/sdb4 1
swift-ring-builder object-1.builder rebalance
swift-ring-builder container.builder create 10 3 1
swift-ring-builder container.builder add r1z1-127.0.0.1:6011/sdb1 1
swift-ring-builder container.builder add r1z2-127.0.0.1:6021/sdb2 1

47
doc/saio/swift/container-reconciler.conf Normal file
View File

@ -0,0 +1,47 @@
[DEFAULT]
# swift_dir = /etc/swift
user = <your-user-name>
# You can specify default log routing here if you want:
# log_name = swift
# log_facility = LOG_LOCAL0
# log_level = INFO
# log_address = /dev/log
#
# comma separated list of functions to call to setup custom log handlers.
# functions get passed: conf, name, log_to_console, log_route, fmt, logger,
# adapted_logger
# log_custom_handlers =
#
# If set, log_udp_host will override log_address
# log_udp_host =
# log_udp_port = 514
#
# You can enable StatsD logging here:
# log_statsd_host = localhost
# log_statsd_port = 8125
# log_statsd_default_sample_rate = 1.0
# log_statsd_sample_rate_factor = 1.0
# log_statsd_metric_prefix =
[container-reconciler]
# reclaim_age = 604800
# interval = 300
# request_tries = 3
[pipeline:main]
pipeline = catch_errors proxy-logging cache proxy-server
[app:proxy-server]
use = egg:swift#proxy
# See proxy-server.conf-sample for options
[filter:cache]
use = egg:swift#memcache
# See proxy-server.conf-sample for options
[filter:proxy-logging]
use = egg:swift#proxy_logging
[filter:catch_errors]
use = egg:swift#catch_errors
# See proxy-server.conf-sample for options

7
doc/saio/swift/swift.conf
View File

@ -2,3 +2,10 @@
# random unique strings that can never change (DO NOT LOSE)
swift_hash_path_prefix = changeme
swift_hash_path_suffix = changeme
[storage-policy:0]
name = gold
default = yes
[storage-policy:1]
name = silver

59
doc/source/admin_guide.rst
View File

@ -2,6 +2,33 @@
Administrator's Guide
=====================
-------------------------
Defining Storage Policies
-------------------------
Defining your Storage Policies is very easy to do with Swift. It is important
that the administrator understand the concepts behind Storage Policies
before actually creating and using them in order to get the most benefit out
of the feature and, more importantly, to avoid having to make unnecessary changes
once a set of policies have been deployed to a cluster.
It is highly recommended that the reader fully read and comprehend
:doc:`overview_policies` before proceeding with administration of
policies. Plan carefully and it is suggested that experimentation be
done first on a non-production cluster to be certain that the desired
configuration meets the needs of the users. See :ref:`upgrade-policy`
before planning the upgrade of your existing deployment.
Following is a high level view of the very few steps it takes to configure
policies once you have decided what you want to do:
#. Define your policies in ``/etc/swift/swift.conf``
#. Create the corresponding object rings
#. Communicate the names of the Storage Policies to cluster users
For a specific example that takes you through these steps, please see
:doc:`policies_saio`
------------------
Managing the Rings
------------------
@ -32,15 +59,15 @@ For more information see :doc:`overview_ring`.
Removing a device from the ring::
swift-ring-builder <builder-file> remove <ip_address>/<device_name>
Removing a server from the ring::
swift-ring-builder <builder-file> remove <ip_address>
Adding devices to the ring:
See :ref:`ring-preparing`
See what devices for a server are in the ring::
swift-ring-builder <builder-file> search <ip_address>
@ -49,7 +76,7 @@ Once you are done with all changes to the ring, the changes need to be
"committed"::
swift-ring-builder <builder-file> rebalance
Once the new rings are built, they should be pushed out to all the servers
in the cluster.
@ -126,7 +153,7 @@ is replaced. Once the drive is replaced, it can be re-added to the ring.
Handling Server Failure
-----------------------
If a server is having hardware issues, it is a good idea to make sure the
If a server is having hardware issues, it is a good idea to make sure the
swift services are not running. This will allow Swift to work around the
failure while you troubleshoot.
@ -149,7 +176,7 @@ Detecting Failed Drives
It has been our experience that when a drive is about to fail, error messages
will spew into `/var/log/kern.log`. There is a script called
`swift-drive-audit` that can be run via cron to watch for bad drives. If
`swift-drive-audit` that can be run via cron to watch for bad drives. If
errors are detected, it will unmount the bad drive, so that Swift can
work around it. The script takes a configuration file with the following
settings:
@ -170,7 +197,7 @@ log_file_pattern /var/log/kern* Location of the log file with globbing
pattern to check against device errors
regex_pattern_X (see below) Regular expression patterns to be used to
locate device blocks with errors in the
log file
log file
================== ============== ===========================================
The default regex pattern used to locate device blocks with errors are
@ -235,7 +262,7 @@ the cluster. Here is an example of a cluster in perfect health::
Queried 2621 containers for dispersion reporting, 19s, 0 retries
100.00% of container copies found (7863 of 7863)
Sample represents 1.00% of the container partition space
Queried 2619 objects for dispersion reporting, 7s, 0 retries
100.00% of object copies found (7857 of 7857)
Sample represents 1.00% of the object partition space
@ -251,7 +278,7 @@ that has::
Queried 2621 containers for dispersion reporting, 8s, 0 retries
100.00% of container copies found (7863 of 7863)
Sample represents 1.00% of the container partition space
Queried 2619 objects for dispersion reporting, 7s, 0 retries
There were 1763 partitions missing one copy.
77.56% of object copies found (6094 of 7857)
@ -285,7 +312,7 @@ You can also run the report for only containers or objects::
100.00% of object copies found (7857 of 7857)
Sample represents 1.00% of the object partition space
Alternatively, the dispersion report can also be output in json format. This
Alternatively, the dispersion report can also be output in json format. This
allows it to be more easily consumed by third party utilities::
$ swift-dispersion-report -j
@ -499,7 +526,7 @@ Request URI Description
This information can also be queried via the swift-recon command line utility::
fhines@ubuntu:~$ swift-recon -h
Usage:
Usage:
usage: swift-recon <server_type> [-v] [--suppress] [-a] [-r] [-u] [-d]
[-l] [--md5] [--auditor] [--updater] [--expirer] [--sockstat]
@ -893,8 +920,8 @@ Metric Name Description
`object-server.PUT.timing` Timing data for each PUT request not resulting in an
error.
`object-server.PUT.<device>.timing` Timing data per kB transferred (ms/kB) for each
non-zero-byte PUT request on each device.
Monitoring problematic devices, higher is bad.
non-zero-byte PUT request on each device.
Monitoring problematic devices, higher is bad.
`object-server.GET.errors.timing` Timing data for GET request errors: bad request,
not mounted, header timestamps before the epoch,
precondition failed.
@ -1046,7 +1073,7 @@ Managing Services
-----------------
Swift services are generally managed with `swift-init`. the general usage is
``swift-init <service> <command>``, where service is the swift service to
``swift-init <service> <command>``, where service is the swift service to
manage (for example object, container, account, proxy) and command is one of:
========== ===============================================
@ -1059,8 +1086,8 @@ shutdown Attempt to gracefully shutdown the service
reload Attempt to gracefully restart the service
========== ===============================================
A graceful shutdown or reload will finish any current requests before
completely stopping the old service. There is also a special case of
A graceful shutdown or reload will finish any current requests before
completely stopping the old service. There is also a special case of
`swift-init all <command>`, which will run the command for all swift services.
In cases where there are multiple configs for a service, a specific config

12
doc/source/container.rst
View File

@ -34,6 +34,18 @@ Container Server
:undoc-members:
:show-inheritance:
.. _container-replicator:
Container Replicator
====================
.. automodule:: swift.container.replicator
:members:
:undoc-members:
:show-inheritance:
.. _container-sync-daemon:
Container Sync
==============

15
doc/source/development_saio.rst
View File

@ -341,6 +341,10 @@ commands are as follows:
.. literalinclude:: /../saio/swift/object-expirer.conf
#. ``/etc/swift/container-reconciler.conf``
.. literalinclude:: /../saio/swift/container-reconciler.conf
#. ``/etc/swift/account-server/1.conf``
.. literalinclude:: /../saio/swift/account-server/1.conf
@ -447,8 +451,15 @@ Setting up scripts for running Swift
.. literalinclude:: /../saio/bin/remakerings
You can expect the ouptut from this command to produce the following::
You can expect the output from this command to produce the following (note
that 2 object rings are created in order to test storage policies in the
SAIO environment however they map to the same nodes)::
Device d0r1z1-127.0.0.1:6010R127.0.0.1:6010/sdb1_"" with 1.0 weight got id 0
Device d1r1z2-127.0.0.1:6020R127.0.0.1:6020/sdb2_"" with 1.0 weight got id 1
Device d2r1z3-127.0.0.1:6030R127.0.0.1:6030/sdb3_"" with 1.0 weight got id 2
Device d3r1z4-127.0.0.1:6040R127.0.0.1:6040/sdb4_"" with 1.0 weight got id 3
Reassigned 1024 (100.00%) partitions. Balance is now 0.00.
Device d0r1z1-127.0.0.1:6010R127.0.0.1:6010/sdb1_"" with 1.0 weight got id 0
Device d1r1z2-127.0.0.1:6020R127.0.0.1:6020/sdb2_"" with 1.0 weight got id 1
Device d2r1z3-127.0.0.1:6030R127.0.0.1:6030/sdb3_"" with 1.0 weight got id 2
@ -465,6 +476,8 @@ Setting up scripts for running Swift
Device d3r1z4-127.0.0.1:6042R127.0.0.1:6042/sdb4_"" with 1.0 weight got id 3
Reassigned 1024 (100.00%) partitions. Balance is now 0.00.
#. Read more about Storage Policies and your SAIO :doc:`policies_saio`
#. Verify the unit tests run::
$HOME/swift/.unittests

2
doc/source/index.rst
View File

@ -45,6 +45,7 @@ Overview and Concepts
Swift's API docs <http://docs.openstack.org/api/openstack-object-storage/1.0/content/>
overview_architecture
overview_ring
overview_policies
overview_reaper
overview_auth
overview_replication
@ -65,6 +66,7 @@ Developer Documentation
development_guidelines
development_saio
policies_saio
development_auth
development_middleware
development_ondisk_backends

2
doc/source/middleware.rst
View File

@ -130,6 +130,8 @@ KeystoneAuth
:members:
:show-inheritance:
.. _list_endpoints:
List Endpoints
==============

9
doc/source/misc.rst
View File

@ -120,3 +120,12 @@ WSGI
.. automodule:: swift.common.wsgi
:members:
:show-inheritance:
.. _storage_policy:
Storage Policy
==============
.. automodule:: swift.common.storage_policy
:members:
:show-inheritance:

33
doc/source/overview_architecture.rst
View File

@ -27,9 +27,9 @@ The Ring
A ring represents a mapping between the names of entities stored on disk and
their physical location. There are separate rings for accounts, containers, and
objects. When other components need to perform any operation on an object,
container, or account, they need to interact with the appropriate ring to
determine its location in the cluster.
one object ring per storage policy. When other components need to perform any
operation on an object, container, or account, they need to interact with the
appropriate ring to determine its location in the cluster.
The Ring maintains this mapping using zones, devices, partitions, and replicas.
Each partition in the ring is replicated, by default, 3 times across the
@ -54,6 +54,33 @@ drives are used in a cluster.
The ring is used by the Proxy server and several background processes
(like replication).
----------------
Storage Policies
----------------
Storage Policies provide a way for object storage providers to differentiate
service levels, features and behaviors of a Swift deployment. Each Storage
Policy configured in Swift is exposed to the client via an abstract name.
Each device in the system is assigned to one or more Storage Policies. This
is accomplished through the use of multiple object rings, where each Storage
Policy has an independent object ring, which may include a subset of hardware
implementing a particular differentiation.
For example, one might have the default policy with 3x replication, and create
a second policy which, when applied to new containers only uses 2x replication.
Another might add SSDs to a set of storage nodes and create a performance tier
storage policy for certain containers to have their objects stored there.
This mapping is then exposed on a per-container basis, where each container
can be assigned a specific storage policy when it is created, which remains in
effect for the lifetime of the container. Applications require minimal
awareness of storage policies to use them; once a container has been created
with a specific policy, all objects stored in it will be done so in accordance
with that policy.
Storage Policies are not implemented as a separate code module but are a core
abstraction of Swift architecture.
-------------
Object Server
-------------

603
doc/source/overview_policies.rst Executable file
View File

@ -0,0 +1,603 @@
================
Storage Policies
================
Storage Policies allow for some level of segmenting the cluster for various
purposes through the creation of multiple object rings. Storage Policies are
not implemented as a separate code module but are an important concept in
understanding Swift architecture.
As described in :doc:`overview_ring`, Swift uses modified hashing rings to
determine where data should reside in the cluster. There is a separate ring
for account databases, container databases, and there is also one object
ring per storage policy. Each object ring behaves exactly the same way
and is maintained in the same manner, but with policies, different devices
can belong to different rings with varying levels of replication. By supporting
multiple object rings, Swift allows the application and/or deployer to
essentially segregate the object storage within a single cluster. There are
many reasons why this might be desirable:
* Different levels of replication: If a provider wants to offer, for example,
2x replication and 3x replication but doesn't want to maintain 2 separate clusters,
they would setup a 2x policy and a 3x policy and assign the nodes to their
respective rings.
* Performance: Just as SSDs can be used as the exclusive members of an account or
database ring, an SSD-only object ring can be created as well and used to
implement a low-latency/high performance policy.
* Collecting nodes into group: Different object rings may have different
physical servers so that objects in specific storage policies are always
placed in a particular data center or geography.
* Different Storage implementations: Another example would be to collect
together a set of nodes that use a different Diskfile (e.g., Kinetic,
GlusterFS) and use a policy to direct traffic just to those nodes.
.. note::
Today, choosing a different storage policy allows the use of different
object rings, but future policies (such as Erasure Coding) will also
change some of the actual code paths when processing a request. Also note
that Diskfile refers to backend object storage plug-in architecture.
-----------------------
Containers and Policies
-----------------------
Policies are implemented at the container level. There are many advantages to
this approach, not the least of which is how easy it makes life on
applications that want to take advantage of them. It also ensures that
Storage Policies remain a core feature of swift independent of the auth
implementation. Policies were not implemented at the account/auth layer
because it would require changes to all auth systems in use by Swift
deployers. Each container has a new special immutable metadata element called
the storage policy index. Note that internally, Swift relies on policy
indexes and not policy names. Policy names exist for human readability and
translation is managed in the proxy. When a container is created, one new
optional header is supported to specify the policy name. If nothing is
specified, the default policy is used (and if no other policies defined,
Policy-0 is considered the default). We will be covering the difference
between default and Policy-0 in the next section.
Policies are assigned when a container is created. Once a container has been
assigned a policy, it cannot be changed until the container is deleted. The implications
on data placement/movement for large datasets would make this a task best left for
applications to perform. Therefore, if a container has an existing policy of,
for example 3x replication, and one wanted to migrate that data to a policy that specifies,
a different replication level, the application would create another container
specifying the other policy name and then simply move the data from one container
to the other. Policies apply on a per container basis allowing for minimal application
awareness; once a container has been created with a specific policy, all objects stored
in it will be done so in accordance with that policy. If a container with a
specific name is deleted (requires the container be empty) a new container may
be created with the same name without any restriction on storage policy
enforced by the deleted container which previously shared the same name.
Containers have a many-to-one relationship with policies meaning that any number
of containers can share one policy. There is no limit to how many containers can use
a specific policy.
The notion of associating a ring with a container introduces an interesting scenario:
What would happen if 2 containers of the same name were created with different
Storage Policies on either side of a network outage at the same time? Furthermore,
what would happen if objects were placed in those containers, a whole bunch of them,
and then later the network outage was restored? Well, without special care it would
be a big problem as an application could end up using the wrong ring to try and find
an object. Luckily there is a solution for this problem, a daemon covered in more
detail later, works tirelessly to identify and rectify this potential scenario.
--------------------
Container Reconciler
--------------------
Because atomicity of container creation cannot be enforced in a
distributed eventually consistent system, object writes into the wrong
storage policy must be eventually merged into the correct storage policy
by an asynchronous daemon. Recovery from object writes during a network
partition which resulted in a split brain container created with
different storage policies are handled by the
`swift-container-reconciler` daemon.
The container reconciler works off a queue similar to the
object-expirer. The queue is populated during container-replication.
It is never considered incorrect to enqueue an object to be evaluated by
the container-reconciler because if there is nothing wrong with the location
of the object the reconciler will simply dequeue it. The
container-reconciler queue is an indexed log for the real location of an
object for which a discrepancy in the storage policy of the container was
discovered.
To determine the correct storage policy of a container, it is necessary
to update the status_changed_at field in the container_stat table when a
container changes status from deleted to re-created. This transaction
log allows the container-replicator to update the correct storage policy
both when replicating a container and handling REPLICATE requests.
Because each object write is a separate distributed transaction it is
not possible to determine the correctness of the storage policy for each
object write with respect to the entire transaction log at a given
container database. As such, container databases will always record the
object write regardless of the storage policy on a per object row basis.
Object byte and count stats are tracked per storage policy in each
container and reconciled using normal object row merge semantics.
The object rows are ensured to be fully durable during replication using
the normal container replication. After the container
replicator pushes its object rows to available primary nodes any
misplaced object rows are bulk loaded into containers based off the
object timestamp under the ".misplaced_objects" system account. The
rows are initially written to a handoff container on the local node, and
at the end of the replication pass the .misplaced_object containers are
replicated to the correct primary nodes.
The container-reconciler processes the .misplaced_objects containers in
descending order and reaps its containers as the objects represented by
the rows are successfully reconciled. The container-reconciler will
always validate the correct storage policy for enqueued objects using
direct container HEAD requests which are accelerated via caching.
Because failure of individual storage nodes in aggregate is assumed to
be common at scale the container-reconciler will make forward progress
with a simple quorum majority. During a combination of failures and
rebalances it is possible that a quorum could provide an incomplete
record of the correct storage policy - so an object write may have to be
applied more than once. Because storage nodes and container databases
will not process writes with an ``X-Timestamp`` less than or equal to
their existing record when objects writes are re-applied their timestamp
is slightly incremented. In order for this increment to be applied
transparently to the client a second vector of time has been added to
Swift for internal use. See :class:`~swift.common.utils.Timestamp`.
As the reconciler applies object writes to the correct storage policy it
cleans up writes which no longer apply to the incorrect storage policy
and removes the rows from the ``.misplaced_objects`` containers. After all
rows have been successfully processed it sleeps and will periodically
check for newly enqueued rows to be discovered during container
replication.
.. _default-policy:
-------------------------
Default versus 'Policy-0'
-------------------------
Storage Policies is a versatile feature intended to support both new and
pre-existing clusters with the same level of flexibility. For that reason, we
introduce the ``Policy-0`` concept which is not the same as the "default"
policy. As you will see when we begin to configure policies, each policy has
both a name (human friendly, configurable) as well as an index (or simply
policy number). Swift reserves index 0 to map to the object ring that's
present in all installations (e.g., ``/etc/swift/object.ring.gz``). You can
name this policy anything you like, and if no policies are defined it will
report itself as ``Policy-0``, however you cannot change the index as there must
always be a policy with index 0.
Another important concept is the default policy which can be any policy
in the cluster. The default policy is the policy that is automatically
chosen when a container creation request is sent without a storage
policy being specified. :ref:`configure-policy` describes how to set the
default policy. The difference from ``Policy-0`` is subtle but
extremely important. ``Policy-0`` is what is used by Swift when
accessing pre-storage-policy containers which won't have a policy - in
this case we would not use the default as it might not have the same
policy as legacy containers. When no other policies are defined, Swift
will always choose ``Policy-0`` as the default.
In other words, default means "create using this policy if nothing else is specified"
and ``Policy-0`` means "use the legacy policy if a container doesn't have one" which
really means use ``object.ring.gz`` for lookups.
.. note::
With the Storage Policy based code, it's not possible to create a
container that doesn't have a policy. If nothing is provided, Swift will
still select the default and assign it to the container. For containers
created before Storage Policies were introduced, the legacy Policy-0 will
be used.
.. _deprecate-policy:
--------------------
Deprecating Policies
--------------------
There will be times when a policy is no longer desired; however simply
deleting the policy and associated rings would be problematic for existing
data. In order to ensure that resources are not orphaned in the cluster (left
on disk but no longer accessible) and to provide proper messaging to
applications when a policy needs to be retired, the notion of deprecation is
used. :ref:`configure-policy` describes how to deprecate a policy.
Swift's behavior with deprecated policies will change as follows:
* The deprecated policy will not appear in /info
* PUT/GET/DELETE/POST/HEAD are still allowed on the pre-existing containers
created with a deprecated policy
* Clients will get an ''400 Bad Request'' error when trying to create a new
container using the deprecated policy
* Clients still have access to policy statistics via HEAD on pre-existing
containers
.. note::
A policy can not be both the default and deprecated. If you deprecate the
default policy, you must specify a new default.
You can also use the deprecated feature to rollout new policies. If you
want to test a new storage policy before making it generally available
you could deprecate the policy when you initially roll it the new
configuration and rings to all nodes. Being deprecated will render it
innate and unable to be used. To test it you will need to create a
container with that storage policy; which will require a single proxy
instance (or a set of proxy-servers which are only internally
accessible) that has been one-off configured with the new policy NOT
marked deprecated. Once the container has been created with the new
storage policy any client authorized to use that container will be able
to add and access data stored in that container in the new storage
policy. When satisfied you can roll out a new ``swift.conf`` which does
not mark the policy as deprecated to all nodes.
.. _configure-policy:
--------------------
Configuring Policies
--------------------
Policies are configured in ``swift.conf`` and it is important that the deployer have a solid
understanding of the semantics for configuring policies. Recall that a policy must have
a corresponding ring file, so configuring a policy is a two-step process. First, edit
your ``/etc/swift/swift.conf`` file to add your new policy and, second, create the
corresponding policy object ring file.
See :doc:`policies_saio` for a step by step guide on adding a policy to the SAIO setup.
Note that each policy has a section starting with ``[storage-policy:N]`` where N is the
policy index. There's no reason other than readability that these be sequential but there
are a number of rules enforced by Swift when parsing this file:
* If a policy with index 0 is not declared and no other policies defined,
Swift will create one
* The policy index must be a non-negative integer
* If no policy is declared as the default and no other policies are
defined, the policy with index 0 is set as the default
* Policy indexes must be unique
* Policy names are required
* Policy names are case insensitive
* Policy names must contain only letters, digits or a dash
* Policy names must be unique
* The policy name 'Policy-0' can only be used for the policy with index 0
* If any policies are defined, exactly one policy must be declared default
* Deprecated policies can not be declared the default
The following is an example of a properly configured ''swift.conf'' file. See :doc:`policies_saio`
for full instructions on setting up an all-in-one with this example configuration.::
[swift-hash]
# random unique strings that can never change (DO NOT LOSE)
swift_hash_path_prefix = changeme
swift_hash_path_suffix = changeme
[storage-policy:0]
name = gold
default = yes
[storage-policy:1]
name = silver
deprecated = yes
Review :ref:`default-policy` and :ref:`deprecate-policy` for more
information about the ``default`` and ``deprecated`` options.
There are some other considerations when managing policies:
* Policy names can be changed (but be sure that users are aware, aliases are
not currently supported but could be implemented in custom middleware!)
* You cannot change the index of a policy once it has been created
* The default policy can be changed at any time, by adding the
default directive to the desired policy section
* Any policy may be deprecated by adding the deprecated directive to
the desired policy section, but a deprecated policy may not also
be declared the default, and you must specify a default - so you
must have policy which is not deprecated at all times.
There will be additional parameters for policies as new features are added
(e.g., Erasure Code), but for now only a section name/index and name are
required. Once ``swift.conf`` is configured for a new policy, a new ring must be
created. The ring tools are not policy name aware so it's critical that the
correct policy index be used when creating the new policy's ring file.
Additional object rings are created in the same manner as the legacy ring
except that '-N' is appended after the word ``object`` where N matches the
policy index used in ``swift.conf``. This naming convention follows the pattern
for per-policy storage node data directories as well. So, to create the ring
for policy 1::
swift-ring-builder object-1.builder create 10 3 1
<and add devices, rebalance using the same naming convention>
.. note::
The same drives can indeed be used for multiple policies and the details
of how that's managed on disk will be covered in a later section, it's
important to understand the implications of such a configuration before
setting one up. Make sure it's really what you want to do, in many cases
it will be, but in others maybe not.
--------------
Using Policies
--------------
Using policies is very simple, a policy is only specified when a container is
initially created, there are no other API changes. Creating a container can
be done without any special policy information::
curl -v -X PUT -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test/myCont0
Which will result in a container created that is associated with the
policy name 'gold' assuming we're using the swift.conf example from
above. It would use 'gold' because it was specified as the default.
Now, when we put an object into this container, it will get placed on
nodes that are part of the ring we created for policy 'gold'.
If we wanted to explicitly state that we wanted policy 'gold' the command
would simply need to include a new header as shown below::
curl -v -X PUT -H 'X-Auth-Token: <your auth token>' \
-H 'X-Storage-Policy: gold' http://127.0.0.1:8080/v1/AUTH_test/myCont1
And that's it! The application does not need to specify the policy name ever
again. There are some illegal operations however:
* If an invalid (typo, non-existent) policy is specified: 400 Bad Request
* if you try to change the policy either via PUT or POST: 409 Conflict
If you'd like to see how the storage in the cluster is being used, simply HEAD
the account and you'll see not only the cumulative numbers, as before, but
per policy statistics as well. In the example below there's 3 objects total
with two of them in policy 'gold' and one in policy 'silver'::
curl -i -X HEAD -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test
and your results will include (some output removed for readability)::
X-Account-Container-Count: 3
X-Account-Object-Count: 3
X-Account-Bytes-Used: 21
X-Storage-Policy-Gold-Object-Count: 2
X-Storage-Policy-Gold-Bytes-Used: 14
X-Storage-Policy-Silver-Object-Count: 1
X-Storage-Policy-Silver-Bytes-Used: 7
--------------
Under the Hood
--------------
Now that we've explained a little about what Policies are and how to
configure/use them, let's explore how Storage Policies fit in at the
nuts-n-bolts level.
Parsing and Configuring
-----------------------
The module, :ref:`storage_policy`, is responsible for parsing the
``swift.conf`` file, validating the input, and creating a global collection of
configured policies via class :class:`.StoragePolicyCollection`. This
collection is made up of policies of class :class:`.StoragePolicy`. The
collection class includes handy functions for getting to a policy either by
name or by index , getting info about the policies, etc. There's also one
very important function, :meth:`~.StoragePolicyCollection.get_object_ring`.
Object rings are now members of the :class:`.StoragePolicy` class and are
actually not instantiated until the :meth:`~.StoragePolicy.load_ring`
method is called. Any caller anywhere in the code base that needs to access
an object ring must use the :data:`.POLICIES` global singleton to access the
:meth:`~.StoragePolicyCollection.get_object_ring` function and provide the
policy index which will call :meth:`~.StoragePolicy.load_ring` if
needed; however, when starting request handling services such as the
:ref:`proxy-server` rings are proactively loaded to provide moderate
protection against a mis-configuration resulting in a run time error. The
global is instantiated when Swift starts and provides a mechanism to patch
policies for the test code.
Middleware
----------
Middleware can take advantage of policies through the :data:`.POLICIES` global
and by importing :func:`.get_container_info` to gain access to the policy
index associated with the container in question. From the index it
can then use the :data:`.POLICIES` singleton to grab the right ring. For example,
:ref:`list_endpoints` is policy aware using the means just described. Another
example is :ref:`recon` which will report the md5 sums for all object rings.
Proxy Server
------------
The :ref:`proxy-server` module's role in Storage Policies is essentially to make sure the
correct ring is used as its member element. Before policies, the one object ring
would be instantiated when the :class:`.Application` class was instantiated and could
be overridden by test code via init parameter. With policies, however, there is
no init parameter and the :class:`.Application` class instead depends on the :data:`.POLICIES`
global singleton to retrieve the ring which is instantiated the first time it's
needed. So, instead of an object ring member of the :class:`.Application` class, there is
an accessor function, :meth:`~.Application.get_object_ring`, that gets the ring from :data:`.POLICIES`.
In general, when any module running on the proxy requires an object ring, it
does so via first getting the policy index from the cached container info. The
exception is during container creation where it uses the policy name from the
request header to look up policy index from the :data:`.POLICIES` global. Once the
proxy has determined the policy index, it can use the :meth:`~.Application.get_object_ring` method
described earlier to gain access to the correct ring. It then has the responsibility
of passing the index information, not the policy name, on to the back-end servers
via the header ``X-Backend-Storage-Policy-Index``. Going the other way, the proxy also
strips the index out of headers that go back to clients, and makes sure they only
see the friendly policy names.
On Disk Storage
---------------
Policies each have their own directories on the back-end servers and are identified by
their storage policy indexes. Organizing the back-end directory structures by policy
index helps keep track of things and also allows for sharing of disks between policies
which may or may not make sense depending on the needs of the provider. More
on this later, but for now be aware of the following directory naming convention:
* ``/objects`` maps to objects associated with Policy-0
* ``/objects-N`` maps to storage policy index #N
* ``/async_pending`` maps to async pending update for Policy-0
* ``/async_pending-N`` maps to async pending update for storage policy index #N
* ``/tmp`` maps to the DiskFile temporary directory for Policy-0
* ``/tmp-N`` maps to the DiskFile temporary directory for policy index #N
* ``/quarantined/objects`` maps to the quarantine directory for Policy-0
* ``/quarantined/objects-N`` maps to the quarantine directory for policy index #N
Note that these directory names are actually owned by the specific Diskfile
Implementation, the names shown above are used by the default Diskfile.
Object Server
-------------
The :ref:`object-server` is not involved with selecting the storage policy
placement directly. However, because of how back-end directory structures are
setup for policies, as described earlier, the object server modules do play a
role. When the object server gets a :class:`.Diskfile`, it passes in the
policy index and leaves the actual directory naming/structure mechanisms to
:class:`.Diskfile`. By passing in the index, the instance of
:class:`.Diskfile` being used will assure that data is properly located in the
tree based on its policy.
For the same reason, the :ref:`object-updater` also is policy aware; as previously
described, different policies use different async pending directories so the
updater needs to know how to scan them appropriately.
The :ref:`object-replicator` is policy aware in that, depending on the policy, it may have to
do drastically different things, or maybe not. For example, the difference in
handling a replication job for 2x versus 3x is trivial; however, the difference in
handling replication between 3x and erasure code is most definitely not. In
fact, the term 'replication' really isn't appropriate for some policies
like erasure code; however, the majority of the framework for collecting and
processing jobs remains the same. Thus, those functions in the replicator are
leveraged for all policies and then there is policy specific code required for
each policy, added when the policy is defined if needed.
The ssync functionality is policy aware for the same reason. Some of the
other modules may not obviously be affected, but the back-end directory
structure owned by :class:`.Diskfile` requires the policy index
parameter. Therefore ssync being policy aware really means passing the
policy index along. See :class:`~swift.obj.ssync_sender` and
:class:`~swift.obj.ssync_receiver` for more information on ssync.
For :class:`.Diskfile` itself, being policy aware is all about managing the back-end
structure using the provided policy index. In other words, callers who get
a :class:`.Diskfile` instance provide a policy index and :class:`.Diskfile`'s job is to keep data
separated via this index (however it chooses) such that policies can share
the same media/nodes if desired. The included implementation of :class:`.Diskfile`
lays out the directory structure described earlier but that's owned within
:class:`.Diskfile`; external modules have no visibility into that detail. A common
function is provided to map various directory names and/or strings
based on their policy index. For example :class:`.Diskfile` defines :func:`.get_data_dir`
which builds off of a generic :func:`.get_policy_string` to consistently build
policy aware strings for various usage.
Container Server
----------------
The :ref:`container-server` plays a very important role in Storage Policies, it is
responsible for handling the assignment of a policy to a container and the
prevention of bad things like changing policies or picking the wrong policy
to use when nothing is specified (recall earlier discussion on Policy-0 versus
default).
The :ref:`container-updater` is policy aware, however its job is very simple, to
pass the policy index along to the :ref:`account-server` via a request header.
The :ref:`container-backend` is responsible for both altering existing DB
schema as well as assuring new DBs are created with a schema that supports
storage policies. The "on-demand" migration of container schemas allows Swift
to upgrade without downtime (sqlite's alter statements are fast regardless of
row count). To support rolling upgrades (and downgrades) the incompatible
schema changes to the ``container_stat`` table are made to a
``container_info`` table, and the ``container_stat`` table is replaced with a
view that includes an ``INSTEAD OF UPDATE`` trigger which makes it behave like
the old table.
The policy index is stored here for use in reporting information
about the container as well as managing split-brain scenario induced
discrepancies between containers and their storage policies. Furthermore,
during split-brain containers must be prepared to track object updates from
multiple policies, so the object table also includes a
``storage_policy_index`` column. Per-policy object counts and bytes are
updated in the ``policy_stat`` table using ``INSERT`` and ``DELETE`` triggers
similar to the pre-policy triggers that updated ``container_stat`` directly.
The :ref:`container-replicator` daemon will pro-actively migrate legacy
schemas as part of its normal consistency checking process when it updates the
``reconciler_sync_point`` entry in the ``container_info`` table. This ensures
that read heavy containers which do not encounter any writes will still get
migrated to be fully compatible with the post-storage-policy queries without
having to fall-back and retry queries with the legacy schema to service
container read requests.
The :ref:`container-sync-daemon` functionality only needs to be policy aware in that it
accesses the object rings. Therefore, it needs to pull the policy index
out of the container information and use it to select the appropriate
object ring from the :data:`.POLICIES` global.
Account Server
--------------
The :ref:`account-server`'s role in Storage Policies is really limited to reporting.
When a HEAD request is made on an account (see example provided earlier),
the account server is provided with the storage policy index and builds
the ``object_count`` and ``byte_count`` information for the client on a per
policy basis.
The account servers are able to report per-storage-policy object and byte
counts because of some policy specific DB schema changes. A policy specific
table, ``policy_stat``, maintains information on a per policy basis (one row
per policy) in the same manner in which the ``account_stat`` table does. The
``account_stat`` table still serves the same purpose and is not replaced by
``policy_stat``, it holds the total account stats whereas ``policy_stat`` just
has the break downs. The backend is also responsible for migrating
pre-storage-policy accounts by altering the DB schema and populating the
``policy_stat`` table for Policy-0 with current ``account_stat`` data at that
point in time.
The per-storage-policy object and byte counts are not updated with each object
PUT and DELETE request container updates to the account server is performed
asynchronously by the ``swift-container-updater``.
.. _upgrade-policy:
Upgrading and Confirming Functionality
--------------------------------------
Upgrading to a version of Swift that has Storage Policy support is not difficult,
in fact, the cluster administrator isn't required to make any special configuration
changes to get going. Swift will automatically begin using the existing object
ring as both the default ring and the Policy-0 ring. Adding the declaration of
policy 0 is totally optional and in its absence, the name given to the implicit
policy 0 will be 'Policy-0'. Let's say for testing purposes that you wanted to take
an existing cluster that already has lots of data on it and upgrade to Swift with
Storage Policies. From there you want to go ahead and create a policy and test a
few things out. All you need to do is:
#. Define your policies in ``/etc/swift/swift.conf``
#. Create the corresponding object rings
#. Create containers and objects and confirm their placement is as expected
For a specific example that takes you through these steps, please see
:doc:`policies_saio`
.. note::
If you downgrade from a Storage Policy enabled version of Swift to an
older version that doesn't support policies, you will not be able to
access any data stored in policies other than the policy with index 0 but
those objects WILL appear in container listings (possibly as duplicates if
there was a network partition and un-reconciled objects). It is EXTREMELY
important that you perform any necessary integration testing on the
upgraded deployment before enabling an additional storage policy to ensure
a consistent API experience for your clients. DO NOT downgrade to a
version of Swift that does not support storage policies once you expose
multiple storage policies.

2
doc/source/overview_replication.rst
View File

@ -93,6 +93,8 @@ systems, it was designed so that around 2% of the hash space on a normal node
will be invalidated per day, which has experimentally given us acceptable
replication speeds.
.. _ssync:
Work continues with a new ssync method where rsync is not used at all and
instead all-Swift code is used to transfer the objects. At first, this ssync
will just strive to emulate the rsync behavior. Once deemed stable it will open

146
doc/source/policies_saio.rst Normal file
View File

@ -0,0 +1,146 @@
===========================================
Adding Storage Policies to an Existing SAIO
===========================================
Depending on when you downloaded your SAIO environment, it may already
be prepared with two storage policies that enable some basic functional
tests. In the event that you are adding a storage policy to an existing
installation, however, the following section will walk you through the
steps for setting up Storage Policies. Note that configuring more than
one storage policy on your development environment is recommended but
optional. Enabling multiple Storage Policies is very easy regardless of
whether you are working with an existing installation or starting a
brand new one.
Now we will create two policies - the first one will be a standard triple
replication policy that we will also explicitly set as the default and
the second will be setup for reduced replication using a factor of 2x.
We will call the first one 'gold' and the second one 'silver'. In this
example both policies map to the same devices because it's also
important for this sample implementation to be simple and easy
to understand and adding a bunch of new devices isn't really required
to implement a usable set of policies.
1. To define your policies, add the following to your ``/etc/swift/swift.conf``
file::
[storage-policy:0]
name = gold
default = yes
[storage-policy:1]
name = silver
See :doc:`overview_policies` for detailed information on ``swift.conf`` policy
options.
2. To create the object ring for the silver policy (index 1), add the following
to your ``bin/remakerings`` script and re-run it (your script may already have
these changes)::
swift-ring-builder object-1.builder create 10 2 1
swift-ring-builder object-1.builder add r1z1-127.0.0.1:6010/sdb1 1
swift-ring-builder object-1.builder add r1z2-127.0.0.1:6020/sdb2 1
swift-ring-builder object-1.builder add r1z3-127.0.0.1:6030/sdb3 1
swift-ring-builder object-1.builder add r1z4-127.0.0.1:6040/sdb4 1
swift-ring-builder object-1.builder rebalance
Note that the reduced replication of the silver policy is only a function
of the replication parameter in the ``swift-ring-builder create`` command
and is not specified in ``/etc/swift/swift.conf``.
3. Copy ``etc/container-reconciler.conf-sample`` to
``/etc/swift/container-reconciler.conf`` and fix the user option::
cp etc/container-reconciler.conf-sample /etc/swift/container-reconciler.conf
sed -i "s/# user.*/user = $USER/g" /etc/swift/container-reconciler.conf
------------------
Using Policies
------------------
Setting up Storage Policies was very simple, and using them is even
simpler. In this section, we will run some commands to create a few
containers with different policies and store objects in them and see how
Storage Policies effect placement of data in Swift.
1. We will be using the list_endpoints middleware to confirm object locations,
so enable that now in your ``proxy-server.conf`` file by adding it to the pipeline
and including the filter section as shown below (be sure to restart your proxy
after making these changes)::
pipeline = catch_errors gatekeeper healthcheck proxy-logging cache bulk \
slo dlo ratelimit crossdomain list-endpoints tempurl tempauth staticweb \
container-quotas account-quotas proxy-logging proxy-server
[filter:list-endpoints]
use = egg:swift#list_endpoints
2. Check to see that your policies are reported via /info::
swift -A http://127.0.0.1:8080/auth/v1.0 -U test:tester -K testing info
You should see this: (only showing the policy output here)::
policies: [{'default': True, 'name': 'gold'}, {'name': 'silver'}]
3. Now create a container without specifying a policy, it will use the
default, 'gold' and then put a test object in it (create the file ``file0.txt``
with your favorite editor with some content)::
curl -v -X PUT -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test/myCont0
curl -X PUT -v -T file0.txt -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test/myCont0/file0.txt
4. Now confirm placement of the object with the :ref:`list_endpoints` middleware::
curl -X GET -v http://127.0.0.1:8080/endpoints/AUTH_test/myCont0/file0.txt
You should see this: (note placement on expected devices)::
["http://127.0.0.1:6030/sdb3/761/AUTH_test/myCont0/file0.txt",
"http://127.0.0.1:6010/sdb1/761/AUTH_test/myCont0/file0.txt",
"http://127.0.0.1:6020/sdb2/761/AUTH_test/myCont0/file0.txt"]
5. Create a container using policy 'silver' and put a different file in it::
curl -v -X PUT -H 'X-Auth-Token: <your auth token>' -H \
"X-Storage-Policy: silver" \
http://127.0.0.1:8080/v1/AUTH_test/myCont1
curl -X PUT -v -T file1.txt -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test/myCont1/
6. Confirm placement of the object for policy 'silver'::
curl -X GET -v http://127.0.0.1:8080/endpoints/AUTH_test/myCont1/file1.txt
You should see this: (note placement on expected devices)::
["http://127.0.0.1:6010/sdb1/32/AUTH_test/myCont1/file1.txt",
"http://127.0.0.1:6040/sdb4/32/AUTH_test/myCont1/file1.txt"]
7. Confirm account information with HEAD, make sure that your container-updater
service is running and has executed once since you performed the PUTs or the
account database won't be updated yet::
curl -i -X HEAD -H 'X-Auth-Token: <your auth token>' \
http://127.0.0.1:8080/v1/AUTH_test
You should see something like this (note that total and per policy stats
object sizes will vary)::
HTTP/1.1 204 No Content
Content-Length: 0
X-Account-Object-Count: 2
X-Account-Bytes-Used: 174
X-Account-Container-Count: 2
X-Account-Storage-Policy-Gold-Object-Count: 1
X-Account-Storage-Policy-Gold-Bytes-Used: 84
X-Account-Storage-Policy-Silver-Object-Count: 1
X-Account-Storage-Policy-Silver-Bytes-Used: 90
X-Timestamp: 1397230339.71525
Content-Type: text/plain; charset=utf-8
Accept-Ranges: bytes
X-Trans-Id: tx96e7496b19bb44abb55a3-0053482c75
Date: Fri, 11 Apr 2014 17:55:01 GMT