swift/doc/source/overview_container_sync.rst
Thiago da Silva 26ff2eb1cb container-sync: Sync static links similar to how we sync SLOs
This allows static symlinks to be synced before their target. Dynamic
symlinks could already be synced even if target object has not been
synced, but static links previously required that target object existed
before it can be PUT. Now, have container_sync middleware plumb in an
override like it does for SLO.

Change-Id: I3bfc62b77b247003adcee6bd4d374168bfd6707d
2020-01-24 17:15:57 -08:00

22 KiB

Container to Container Synchronization

Overview

Swift has a feature where all the contents of a container can be mirrored to another container through background synchronization. Swift cluster operators configure their cluster to allow/accept sync requests to/from other clusters, and the user specifies where to sync their container to along with a secret synchronization key.

Note

If you are using the Large Objects <large-objects> feature and syncing to another cluster then you will need to ensure that manifest files and segment files are synced. If segment files are in a different container than their manifest then both the manifest's container and the segments' container must be synced. The target container for synced segment files must always have the same name as their source container in order for them to be resolved by synced manifests.

Be aware that manifest files may be synced before segment files even if they are in the same container and were created after the segment files.

In the case of Static Large Objects <static-large-objects>, a GET request for a manifest whose segments have yet to be completely synced will fail with none or only part of the large object content being returned.

In the case of Dynamic Large Objects <dynamic-large-objects>, a GET request for a manifest whose segments have yet to be completely synced will either fail or return unexpected (and most likely incorrect) content.

Note

If you are using encryption middleware in the cluster from which objects are being synced, then you should follow the instructions for container_sync_client_config to be compatible with encryption.

Note

If you are using symlink middleware in the cluster from which objects are being synced, then you should follow the instructions for symlink_container_sync_client_config to be compatible with symlinks.

Be aware that symlinks may be synced before their targets even if they are in the same container and were created after the target objects. In such cases, a GET for the symlink will fail with a 404 Not Found error. If the target has been overwritten, a GET may produce an older version (for dynamic links) or a 409 Conflict error (for static links).

Configuring Container Sync

Create a container-sync-realms.conf file specifying the allowable clusters and their information:

[realm1]
key = realm1key
key2 = realm1key2
cluster_clustername1 = https://host1/v1/
cluster_clustername2 = https://host2/v1/

[realm2]
key = realm2key
key2 = realm2key2
cluster_clustername3 = https://host3/v1/
cluster_clustername4 = https://host4/v1/

Each section name is the name of a sync realm. A sync realm is a set of clusters that have agreed to allow container syncing with each other. Realm names will be considered case insensitive.

key is the overall cluster-to-cluster key used in combination with the external users' key that they set on their containers' X-Container-Sync-Key metadata header values. These keys will be used to sign each request the container sync daemon makes and used to validate each incoming container sync request.

key2 is optional and is an additional key incoming requests will be checked against. This is so you can rotate keys if you wish; you move the existing key to key2 and make a new key value.

Any values in the realm section whose names begin with cluster_ will indicate the name and endpoint of a cluster and will be used by external users in their containers' X-Container-Sync-To metadata header values with the format //realm_name/cluster_name/account_name/container_name. Realm and cluster names are considered case insensitive.

The endpoint is what the container sync daemon will use when sending out requests to that cluster. Keep in mind this endpoint must be reachable by all container servers, since that is where the container sync daemon runs. Note that the endpoint ends with /v1/ and that the container sync daemon will then add the account/container/obj name after that.

Distribute this container-sync-realms.conf file to all your proxy servers and container servers.

You also need to add the container_sync middleware to your proxy pipeline. It needs to be after any memcache middleware and before any auth middleware. The [filter:container_sync] section only needs the use item. For example:

[pipeline:main]
pipeline = healthcheck proxy-logging cache container_sync tempauth proxy-logging proxy-server

[filter:container_sync]
use = egg:swift#container_sync

The container sync daemon will use an internal client to sync objects. Even if you don't configure the internal client, the container sync daemon will work with default configuration. The default configuration is the same as internal-client.conf-sample. If you want to configure the internal client, please update internal_client_conf_path in container-server.conf. The configuration file at the path will be used for the internal client.

Old-Style: Configuring a Cluster's Allowable Sync Hosts

This section is for the old-style of using container sync. See the previous section, Configuring Container Sync, for the new-style.

With the old-style, the Swift cluster operator must allow synchronization with a set of hosts before the user can enable container synchronization. First, the backend container server needs to be given this list of hosts in the container-server.conf file:

[DEFAULT]
# This is a comma separated list of hosts allowed in the
# X-Container-Sync-To field for containers.
# allowed_sync_hosts = 127.0.0.1
allowed_sync_hosts = host1,host2,etc.
...

[container-sync]
# You can override the default log routing for this app here (don't
# use set!):
# log_name = container-sync
# log_facility = LOG_LOCAL0
# log_level = INFO
# Will sync, at most, each container once per interval
# interval = 300
# Maximum amount of time to spend syncing each container
# container_time = 60

Logging Container Sync

Currently, log processing is the only way to track sync progress, problems, and even just general activity for container synchronization. In that light, you may wish to set the above log_ options to direct the container-sync logs to a different file for easier monitoring. Additionally, it should be noted there is no way for an end user to monitor sync progress or detect problems other than HEADing both containers and comparing the overall information.

Container Sync Statistics

Container Sync INFO level logs contain activity metrics and accounting information for insightful tracking. Currently two different statistics are collected:

About once an hour or so, accumulated statistics of all operations performed by Container Sync are reported to the log file with the following format:

Since (time): (sync) synced [(delete) deletes, (put) puts], (skip) skipped, (fail) failed
time

last report time

sync

number of containers with sync turned on that were successfully synced

delete

number of successful DELETE object requests to the target cluster

put

number of successful PUT object request to the target cluster

skip

number of containers whose sync has been turned off, but are not yet cleared from the sync store

fail

number of containers with failure (due to exception, timeout or other reason)

For each container synced, per container statistics are reported with the following format:

Container sync report: (container), time window start: (start), time window end: %(end), puts: (puts), posts: (posts), deletes: (deletes), bytes: (bytes), sync_point1: (point1), sync_point2: (point2), total_rows: (total)
container

account/container statistics are for

start

report start time

end

report end time

puts

number of successful PUT object requests to the target container

posts

N/A (0)

deletes

number of successful DELETE object requests to the target container

bytes

number of bytes sent over the network to the target container

point1

progress indication - the container's x_container_sync_point1

point2

progress indication - the container's x_container_sync_point2

total

number of objects processed at the container

It is possible that more than one server syncs a container, therefore log files from all servers need to be evaluated

Using the swift tool to set up synchronized containers

Note

The swift tool is available from the python-swiftclient library.

Note

You must be the account admin on the account to set synchronization targets and keys.

You simply tell each container where to sync to and give it a secret synchronization key. First, let's get the account details for our two cluster accounts:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing stat -v
StorageURL: http://cluster1/v1/AUTH_208d1854-e475-4500-b315-81de645d060e
Auth Token: AUTH_tkd5359e46ff9e419fa193dbd367f3cd19
   Account: AUTH_208d1854-e475-4500-b315-81de645d060e
Containers: 0
   Objects: 0
     Bytes: 0

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 stat -v
StorageURL: http://cluster2/v1/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c
Auth Token: AUTH_tk816a1aaf403c49adb92ecfca2f88e430
   Account: AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c
Containers: 0
   Objects: 0
     Bytes: 0

Now, let's make our first container and tell it to synchronize to a second we'll make next:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing post \
  -t '//realm_name/clustername2/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c/container2' \
  -k 'secret' container1

The -t indicates the cluster to sync to, which is the realm name of the section from container-sync-realms.conf, followed by the cluster name from that section (without the cluster_ prefix), followed by the account and container names we want to sync to. The -k specifies the secret key the two containers will share for synchronization; this is the user key, the cluster key in container-sync-realms.conf will also be used behind the scenes.

Now, we'll do something similar for the second cluster's container:

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 post \
  -t '//realm_name/clustername1/AUTH_208d1854-e475-4500-b315-81de645d060e/container1' \
  -k 'secret' container2

That's it. Now we can upload a bunch of stuff to the first container and watch as it gets synchronized over to the second:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing \
  upload container1 .
photo002.png
photo004.png
photo001.png
photo003.png

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 \
  list container2

[Nothing there yet, so we wait a bit...]

Note

If you're an operator running saio and just testing, each time you configure a container for synchronization and place objects in the source container you will need to ensure that container-sync runs before attempting to retrieve objects from the target container. That is, you need to run:

swift-init container-sync once

Now expect to see objects copied from the first container to the second:

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 \
  list container2
photo001.png
photo002.png
photo003.png
photo004.png

You can also set up a chain of synced containers if you want more than two. You'd point 1 -> 2, then 2 -> 3, and finally 3 -> 1 for three containers. They'd all need to share the same secret synchronization key.

Using curl (or other tools) instead

So what's swift doing behind the scenes? Nothing overly complicated. It translates the -t <value> option into an X-Container-Sync-To: <value> header and the -k <value> option into an X-Container-Sync-Key: <value> header.

For instance, when we created the first container above and told it to synchronize to the second, we could have used this curl command:

$ curl -i -X POST -H 'X-Auth-Token: AUTH_tkd5359e46ff9e419fa193dbd367f3cd19' \
  -H 'X-Container-Sync-To: //realm_name/clustername2/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c/container2' \
  -H 'X-Container-Sync-Key: secret' \
  'http://cluster1/v1/AUTH_208d1854-e475-4500-b315-81de645d060e/container1'
HTTP/1.1 204 No Content
Content-Length: 0
Content-Type: text/plain; charset=UTF-8
Date: Thu, 24 Feb 2011 22:39:14 GMT

Old-Style: Using the swift tool to set up synchronized containers

Note

The swift tool is available from the python-swiftclient library.

Note

You must be the account admin on the account to set synchronization targets and keys.

This is for the old-style of container syncing using allowed_sync_hosts.

You simply tell each container where to sync to and give it a secret synchronization key. First, let's get the account details for our two cluster accounts:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing stat -v
StorageURL: http://cluster1/v1/AUTH_208d1854-e475-4500-b315-81de645d060e
Auth Token: AUTH_tkd5359e46ff9e419fa193dbd367f3cd19
   Account: AUTH_208d1854-e475-4500-b315-81de645d060e
Containers: 0
   Objects: 0
     Bytes: 0

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 stat -v
StorageURL: http://cluster2/v1/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c
Auth Token: AUTH_tk816a1aaf403c49adb92ecfca2f88e430
   Account: AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c
Containers: 0
   Objects: 0
     Bytes: 0

Now, let's make our first container and tell it to synchronize to a second we'll make next:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing post \
  -t 'http://cluster2/v1/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c/container2' \
  -k 'secret' container1

The -t indicates the URL to sync to, which is the StorageURL from cluster2 we retrieved above plus the container name. The -k specifies the secret key the two containers will share for synchronization. Now, we'll do something similar for the second cluster's container:

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 post \
  -t 'http://cluster1/v1/AUTH_208d1854-e475-4500-b315-81de645d060e/container1' \
  -k 'secret' container2

That's it. Now we can upload a bunch of stuff to the first container and watch as it gets synchronized over to the second:

$ swift -A http://cluster1/auth/v1.0 -U test:tester -K testing \
  upload container1 .
photo002.png
photo004.png
photo001.png
photo003.png

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 \
  list container2

[Nothing there yet, so we wait a bit...]
[If you're an operator running SAIO and just testing, you may need to
 run 'swift-init container-sync once' to perform a sync scan.]

$ swift -A http://cluster2/auth/v1.0 -U test2:tester2 -K testing2 \
  list container2
photo001.png
photo002.png
photo003.png
photo004.png

You can also set up a chain of synced containers if you want more than two. You'd point 1 -> 2, then 2 -> 3, and finally 3 -> 1 for three containers. They'd all need to share the same secret synchronization key.

Old-Style: Using curl (or other tools) instead

This is for the old-style of container syncing using allowed_sync_hosts.

So what's swift doing behind the scenes? Nothing overly complicated. It translates the -t <value> option into an X-Container-Sync-To: <value> header and the -k <value> option into an X-Container-Sync-Key: <value> header.

For instance, when we created the first container above and told it to synchronize to the second, we could have used this curl command:

$ curl -i -X POST -H 'X-Auth-Token: AUTH_tkd5359e46ff9e419fa193dbd367f3cd19' \
  -H 'X-Container-Sync-To: http://cluster2/v1/AUTH_33cdcad8-09fb-4940-90da-0f00cbf21c7c/container2' \
  -H 'X-Container-Sync-Key: secret' \
  'http://cluster1/v1/AUTH_208d1854-e475-4500-b315-81de645d060e/container1'
HTTP/1.1 204 No Content
Content-Length: 0
Content-Type: text/plain; charset=UTF-8
Date: Thu, 24 Feb 2011 22:39:14 GMT

What's going on behind the scenes, in the cluster?

Container ring devices have a directory called containers, where container databases reside. In addition to containers, each container ring device also has a directory called sync-containers. sync-containers holds symlinks to container databases that were configured for container sync using x-container-sync-to and x-container-sync-key metadata keys.

The swift-container-sync process does the job of sending updates to the remote container. This is done by scanning sync-containers for container databases. For each container db found, newer rows since the last sync will trigger PUTs or DELETEs to the other container.

sync-containers is maintained as follows: Whenever the container-server processes a PUT or a POST request that carries x-container-sync-to and x-container-sync-key metadata keys the server creates a symlink to the container database in sync-containers. Whenever the container server deletes a synced container, the appropriate symlink is deleted from sync-containers.

In addition to the container-server, the container-replicator process does the job of identifying containers that should be synchronized. This is done by scanning the local devices for container databases and checking for x-container-sync-to and x-container-sync-key metadata values. If they exist then a symlink to the container database is created in a sync-containers sub-directory on the same device.

Similarly, when the container sync metadata keys are deleted, the container server and container-replicator would take care of deleting the symlinks from sync-containers.

Note

The swift-container-sync process runs on each container server in the cluster and talks to the proxy servers (or load balancers) in the remote cluster. Therefore, the container servers must be permitted to initiate outbound connections to the remote proxy servers (or load balancers).

The actual syncing is slightly more complicated to make use of the three (or number-of-replicas) main nodes for a container without each trying to do the exact same work but also without missing work if one node happens to be down.

Two sync points are kept in each container database. When syncing a container, the container-sync process figures out which replica of the container it has. In a standard 3-replica scenario, the process will have either replica number 0, 1, or 2. This is used to figure out which rows belong to this sync process and which ones don't.

An example may help. Assume a replica count of 3 and database row IDs are 1..6. Also, assume that container-sync is running on this container for the first time, hence SP1 = SP2 = -1. :

SP1
SP2
 |
 v
-1 0 1 2 3 4 5 6

First, the container-sync process looks for rows with id between SP1 and SP2. Since this is the first run, SP1 = SP2 = -1, and there aren't any such rows. :

SP1
SP2
 |
 v
-1 0 1 2 3 4 5 6

Second, the container-sync process looks for rows with id greater than SP1, and syncs those rows which it owns. Ownership is based on the hash of the object name, so it's not always guaranteed to be exactly one out of every three rows, but it usually gets close. For the sake of example, let's say that this process ends up owning rows 2 and 5.

Once it's finished trying to sync those rows, it updates SP1 to be the biggest row-id that it's seen, which is 6 in this example. :

SP2           SP1
 |             |
 v             v
-1 0 1 2 3 4 5 6

While all that was going on, clients uploaded new objects into the container, creating new rows in the database. :

SP2           SP1
 |             |
 v             v
-1 0 1 2 3 4 5 6 7 8 9 10 11 12

On the next run, the container-sync starts off looking at rows with ids between SP1 and SP2. This time, there are a bunch of them. The sync process try to sync all of them. If it succeeds, it will set SP2 to equal SP1. If it fails, it will set SP2 to the failed object and will continue to try all other objects till SP1, setting SP2 to the first object that failed.

Under normal circumstances, the container-sync processes will have already taken care of synchronizing all rows, between SP1 and SP2, resulting in a set of quick checks. However, if one of the sync processes failed for some reason, then this is a vital fallback to make sure all the objects in the container get synchronized. Without this seemingly-redundant work, any container-sync failure results in unsynchronized objects. Note that the container sync will persistently retry to sync any faulty object until success, while logging each failure.

Once it's done with the fallback rows, and assuming no faults occurred, SP2 is advanced to SP1. :

SP2
SP1
 |
 v
-1 0 1 2 3 4 5 6 7 8 9 10 11 12

Then, rows with row ID greater than SP1 are synchronized (provided this container-sync process is responsible for them), and SP1 is moved up to the greatest row ID seen. :

SP2            SP1
 |              |
 v              v
-1 0 1 2 3 4 5 6 7 8 9 10 11 12