Skip to main content
Version: 8.2

Backup and restore Operate and Tasklist data

note

This release introduces breaking changes, including:

  • The get backup state API and response codes.
  • The utilized URL has changed. For example, curl 'http://localhost:8080/actuator/backups' rather than the previously used backup.
  • backupId must be of integer type now instead of string, which is in sync with Zeebe backupId requirements.

Operate stores its data over multiple indices in Elasticsearch. Backup of Operate data includes several Elasticsearch snapshots containing sets of Operate indices. Each backup is identified by backupId. For example, a backup with an id of 123 may contain the following Elasticsearch snapshots:

camunda_operate_123_8.1.0_part_1_of_6
camunda_operate_123_8.1.0_part_2_of_6
camunda_operate_123_8.1.0_part_3_of_6
camunda_operate_123_8.1.0_part_4_of_6
camunda_operate_123_8.1.0_part_5_of_6
camunda_operate_123_8.1.0_part_6_of_6

Operate provides an API to perform a backup and manage backups (list, check state, delete). Restore a backup using the standard Elasticsearch API.

note

The backup API can be reached via the Actuator management port, which by default is the same as application HTTP port (and in turn defaults to 8080). The port may be reconfigured with the help of management.server.port configuration parameter.

Prerequisites

Before you can use the backup and restore feature:

  1. The Elasticsearch repository must be configured.
  2. Operate and Tasklist must be configured with the repository name using the following configuration parameters:
for Operate:
camunda.operate: backup.repositoryName=<repository name>

for Tasklist:
camunda.tasklist: backup.repositoryName=<repository name>

or with environmental variables:

for Operate:
CAMUNDA_OPERATE_BACKUP_REPOSITORYNAME=<repository name>

for Tasklist:
CAMUNDA_TASKLIST_BACKUP_REPOSITORYNAME=<repository name>

Create backup API

During backup creation Operate can continue running. To create the backup, call the following endpoint:

POST actuator/backups
{
"backupId": <backupId>
}

Response:

CodeDescription
200 OKBackup was successfully started, snapshots will be created asynchronously. List of snapshots is returned in the response body (see example below). This list must be persisted together with the backup id to be able to restore it later.
400 Bad RequestIn case something is wrong with backupId, e.g. the same backup id already exists.
500 Server ErrorAll other errors, e.g. ES returned error response when attempting to create a snapshot.
502 Bad GatewayElasticsearch is not accessible, the request can be retried when it is back.

Example request:

curl --request POST 'http://localhost:8080/actuator/backups' \
-H 'Content-Type: application/json' \
-d '{ "backupId": 123 }'

Example response:

{
"scheduledSnapshots": [
"camunda_operate_123_8.2.0_part_1_of_6",
"camunda_operate_123_8.2.0_part_2_of_6",
"camunda_operate_123_8.2.0_part_3_of_6",
"camunda_operate_123_8.2.0_part_4_of_6",
"camunda_operate_123_8.2.0_part_5_of_6",
"camunda_operate_123_8.2.0_part_6_of_6"
]
}

Get backup state API

As a backup is created asynchronously, call the following endpoint to check the state of the backup:

GET actuator/backups/{backupId}

Response:

CodeDescription
200 OKBackup state could be determined and is returned in the response body.
404 Not FoundBackup with given id does not exist.
500 Server ErrorAll other errors, e.g. ES returned error response when attempting to execute the query.
502 Bad GatewayElasticsearch is not accessible, the request can be retried when it is back.

For example, the request could look like this:

curl 'http://localhost:8080/actuator/backups/123'

Example response:

{
"backupId": 123,
"state": "COMPLETED",
"failureReason": null,
"details": [
//here goes the list of all Elasticsearch snapshots included in the backup
{
"snapshotName": "camunda_operate_123_8.2.0_part_1_of_6",
"state": "SUCCESS",
"startTime": "2023-01-01T10:10:10.100+0000",
"failures": []
},
<..>
]
}

Possible states of the backup:

  • COMPLETED: Backup can be used for restoring the data.
  • IN_PROGRESS: Wait until the backup completes to use it for restore.
  • FAILED: Something went wrong when creating this backup. To find out the exact problem, use the Elasticsearch get snapshot status API for each of the snapshots included in the given backup.
  • INCOMPATIBLE: Backup is incompatible with the current Elasticsearch version.
  • INCOMPLETE: Backup is incomplete (e.g. when backup process was interrupted).

State of the individual snapshot is a copy of Elasticsearch state.

Get backups list API

To get the list of existing backups, the following endpoint can be used:

GET actuator/backups

Response:

CodeDescription
200 OKBackup list could be determined and is returned in the response body. Can be an empty response in case no backups were created yet.
404 Not FoundBackup repository is not configured.
500 Server ErrorAll other errors, e.g. ES returned error response when attempting to execute the query.
502 Bad GatewayElasticsearch is not accessible, the request can be retried when it is back.

For example, the request could look like this:

curl 'http://localhost:8080/actuator/backups'

Response will contain JSON with array of objects representing state of each backup (see get backup state API endpoint).

Delete backup API

To delete all the Elasticsearch snapshots associated with the specific backup id, the following endpoint may be used:

DELETE actuator/backups/123

Response:

CodeDescription
204 No ContentAll commands to delete corresponding ELS snapshots were successfully sent to ELS. ELS will continue deletion asynchronously.
404 Not FoundNot a single snapshot corresponding to given ID exist.
500 Server ErrorAll other errors, e.g. ES returned error response when attempting to execute the query.
502 Bad GatewayElasticsearch is not accessible, the request can be retried when it is back.

Restore backup

There is no Operate API to preform the backup restore. Instead, use the Elasticsearch restore snapshot API.

note

Operate must not be running while a backup restore is taking place.

To restore the backup with a known backup id, you must restore all the snapshots this backup contains (check the response of the create backup API).

Example of Elasticsearch query:

curl --request POST `http://localhost:9200/_snapshot/test/camunda_operate_123_8.1.0-snapshot_part_1_of_6/_restore?wait_for_completion=true`

To summarize, the process may look as follows:

  1. Stop Operate.
  2. Ensure there are no Operate indices present in Elasticsearch (otherwise the restore process will fail).
  3. Iterate over all Elasticsearch snapshots included in the desired backup and restore them using the Elasticsearch restore snapshot API.
  4. Start Operate.

Backup and restore of Tasklist data

Backup and restore of Tasklist may be performed in exactly the same way as Operate data.