A newer version of this documentation is available.

View Latest

Manage Inter-Sync Gateway Replications

      +

      Managing inter-Sync Gateway replications

      Related topics: Overview | Run | Manage | Monitor | Conflict

      Context Clarification

      This content relates only to inter-Sync Gateway replication in Sync Gateway 2.8+. For documentation on pre-2.8 inter-Sync Gateway replication (also known as SG Replicate) — see the documentation for the appropriate release.

      Admin capabilities

      The Admin REST API provides two endpoints to assist in the monitoring, administration and management of replications. These enable users to examine, update, start and stop replications:

      You can use the endpoints manually or by using automation (for example, scripts or a container orchestration system such as Kubernetes).

      The available endpoints used for admin tasks are:

      • _replication — Retrieve, Update or Remove a replication definition

      • _replicationStatus — Stop, Start or Reset a replication

      Replications always run on the node on which they are configured. Users can only access replications on the node from which they make the request.

      Getting Replication Details

      You can view the current definition details of a replication. This includes replications configured in the JSON file and those initialized using the Admin REST API. You can do this for:

      • Individual replications (Example 1)

      • All replications defined for a specified database (Example 2)

      Replication information is returned a JSON object.

      Example 1. Get a replication definition
      • Request

      • Response

      curl --location --request GET 'http://localhost:4985/db1-local/_replication/db1-rep-id1' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        Payload in body
      {
        "db1-rep-id1": {
          "replication_id": "db1-rep-id1",
          "remote": "http://user1:****@example.com:4984/db1-remote",
          "direction": "pull",
          "purge_on_removal": true,
          "continuous": true,
          "filter": "sync_gateway/bychannel",
          "query_params": {
            "channels": [
              "channel.user1"
            ]
          },
          "assigned_node": "1de4994d136b982e"
        }
      }

      The following example retrieves definitions for all replications on a specified database, regardless of the node on which it was configured. The results are returned in an array; one entry per replication.

      Example 2. Get all replication definitions (for a database)
      • Request

      • Response

      curl --location --request GET 'http://localhost:4985/db1-local/_replication' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        Payload in body
      {
        "db1-rep-id1": {
            "replication_id": "db1-rep-id1",
            "remote": "http://user1:****@example1.com:4984/db1-remote",
            "direction": "pushAndPull",
            "conflict_resolution_type": "remoteWins",
            "purge_on_removal": true,
            "enable_delta_sync": true,
            "initial_state": "stopped",
            "continuous": true,
            "filter": "sync_gateway/bychannel",
            "query_params": {
                "channels": [
                    "channel.user1"
                ]
            },
            "batch_size": 1000,
            "assigned_node": "2c9b0d00a4e7c65a"
        },
        "db1-rep-id2": {
            "replication_id": "db1-rep-id2",
            "remote": "http://user1:****@example2.com:4984/db1-remote",
            "direction": "pushAndPull",
            "conflict_resolution_type": "remoteWins",
            "purge_on_removal": true,
            "enable_delta_sync": true,
            "max_backoff_time": 5,
            "initial_state": "running",
            "continuous": true,
            "filter": "sync_gateway/bychannel",
            "query_params": {
                "channels": [
                    "channel.user1"
                ]
            },
            "adhoc": true,
            "batch_size": 1000,
            "assigned_node": "2c9b0d00a4e7c65a"
        }
      }

      Updating a Replication

      You can update an existing replication’s definition, whether configured or initialized by Admin REST API, by providing the details you want to change in an API call (Example 3). Changes will only be made to those parameters provided in the call.

      If you change the remote URI it must be to a valid URI.

      How do I change an existing replication’s definition details?

      Send a PUT request to the _replication endpoint. Specify just the changed items in the JSON body.

      Example 3. Update a replication’s details
      • Request

      • Response

      curl --location --request PUT 'http://localhost:4985/db1-local/_replication/db1-rep-id1 \
      --header 'Content-Type: application/json' \
      --data-raw '{
        "direction": "push",
        "purge_on_removal":false, // set back to default
        "remote": "http://user1:password1@example.com:4984/db1-remote",
        "filter":"sync_gateway/bychannel",
        "query_params": {
          "channels":["channel.user1"]
        },
        "continuous": false
        }'

      A successful update will return a 200 response, with the following body:

      Success Response::
        Response Status 200 OK
        No payload

      If the replication_id in the body does not match that quoted in the URI you will see a 400 response as below.

      Bad Request Response::
        Response Status 400 Bad request
      
      {
          "error": "Bad Request",
          "reason": "Replication ID in body \"db1-rep-id1\" does not match request URI"
      }

      Removing a Replication

      Removing a replication will delete:

      • The persisted replication definition

      • All checkpointsglossary icon associated with the replication

      • All replication status information associated with the replication

      To find the replication_id of an existing replication see Getting Replication Status Data.

      Action: Send a DELETE request to the replication endpoint specifying the replication_id to remove

      Example 4. Removing a replication
      • Request

      • Response

      curl --location --request DELETE 'http://localhost:4985/db2-local/_replication/db2-rep-id3' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        No payload

      Getting Replication Status Data

      Sync Gateway provides easy access to replication status data through the Admin REST API.

      You can obtain the replication status details for a specific replication, or for all replications across all nodes. This option can be useful, for example, to find any auto-generated replication_id details needed to enable further replication management activities.

      Replications always run on the node on which they are configured. Users can only access replications on the node from which they make the request.

      For more information on monitoring see: Inter Sync Gateway Sync - Monitor

      Example 5. For a Single Replication

      This example targets a known replication-id and returns its status data.

      • Request

      • Response

      curl --location --request GET 'http://localhost:4985/db1/_replicationStatus/db1-rep-id2' \
      --header 'Content-Type: application/json' \
      [
        {
          "replication_id": "db1-rep-id2",
          "docs_read": 0,
          "docs_written": 10,
          "doc_write_failures": 0,
          "doc_write_conflict": 0,
          "status": "running",
          "rejected_by_remote": 0,
          "rejected_by_local": 0,
          "last_seq_pull": "8851",
          "last_seq_push": "10402"
      }
      ]
      Example 6. For All Replications

      This example targets all replications across all nodes. It filters the results using a query string — see: Inter Sync Gateway Sync - Monitor for more on using this option.

      • Request

      • Response

      curl --location --request GET "http://localhost:4985/db1-local/_replicationStatus?activeOnly=false&includeConfig=true&localOnly=false&includeError=true" \ (1)
      --header 'Content-Type: application/json' \
      1 This example’s criteria selects replications with any status (including errors), on local and remote nodes. The returned status details also include replication definition details.
      [
        {
          "replication_id": "db1-rep-id1-pull",
          "docs_read": 0,
          "docs_written": 0,
          "doc_write_failures": 0,
          "doc_write_conflict": 0,
          "status": "running",
          "rejected_by_remote": 0,
          "rejected_by_local": 0,
          "config": { (1)
              "replication_id": "db1-rep-id1-pull",
              "cancel": true,
              "direction": "pull",
              "purge-on-removal": true,
              "remote": "http://user:****@example.com:4985/db1-remote",
              "filter":"sync_gateway/bychannel",
              "query_params": {
                "channels": ["channel1.user1"]
              },
              "continuous": true
          }
        },
        {
          "replication_id": "db1-rep-id2",
          "docs_read": 0,
          "docs_written": 0,
          "doc_write_failures": 0,
          "doc_write_conflict": 0,
          "status": "stopped",  (2)
          "rejected_by_remote": 0,
          "rejected_by_local": 0,
          "config": {
              "replication_id": "db1-rep-id2",
              "direction": "pull",
              "remote": "http://user:****@example.com:4985/db1-remote",
              "continuous": true
            }
        },
        {
          "replication_id": "db2-rep-id1",
          "docs_read": 0,
          "docs_written": 0,
          "doc_write_failures": 0,
          "doc_write_conflict": 0,
          "status": "error", (3)
          "rejected_by_remote": 0,
          "rejected_by_local": 0,
          "config": {
            "replication_id": "db2-rep-id1",
            "direction": "pull",
            "remote": "http://user:****@example2.com:4985/db2-remote",
            "continuous": true
          }
        }
      ]
      1 The configuration details included because includeConfig=true
      2 "Stopped" replications included because activeOnly=false
      3 "error" replications included because includeError=true

      Starting a Replication

      You can start a persistent or ad hoc replication not already in the running state. You need to specify the replication_id.

      If the replication is resetting it cannot be started until the reset is complete.

      Action: Send a POST request to the _replicationStatus endpoint with action=start

      Example 7. Start a replication
      • Request

      • Response

      curl --location --request PUT 'http://localhost:4985/db1-local/_replicationStatus/\{{db1-rep-id}}?action=start' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        Payload in body
      {
        "replication_id": "db1-rep-id1",
        "docs_read": 0,
        "docs_written": 10,
        "doc_write_failures": 0,
        "doc_write_conflict": 0,
        "status": "running",
        "rejected_by_remote": 0,
        "rejected_by_local": 0,
        "last_seq_pull": "8851",
        "last_seq_push": "10402"
      }

      Stopping a Replication

      You can stop a persistent or ad hoc replication not already in the stopped state. You can use this, for example, to offline an edge cluster without waiting for a long replication to complete.

      Action: Send a POST request to the _replicationStatus endpoint with action=stop

      Example 8. Stopping replications
      • Request

      • Response

      curl --location --request PUT 'http://localhost:4985/db1-local/_replicationStatus/\{{db1-rep-id1}}?action=stop' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        Payload in body
      {
        "replication_id": "db1-rep-id1",
        "docs_read": 0,
        "docs_written": 0,
        "doc_write_failures": 0,
        "doc_write_conflict": 0,
        "status": "stopped",
        "rejected_by_remote": 0,
        "rejected_by_local": 0
      }

      Resetting a Replication

      You can reset a persistent replication not in the running state. This can be useful to escape a system state where one or more documents have failed to sync but where resuming from previous synced checkpointglossary icon would skip over those documents. You need to specify the replication_id.

      If the replication is resetting it cannot be started until the reset is complete. The replication must be stopped before it can be reset.

      Action: Send a POST request to the _replicationStatus endpoint with action=reset

      Example 9. Reset a replication
      • Request

      • Response

      curl --location --request PUT 'http://localhost:4985/db1-local/_replicationStatus/\{{db1-rep-id2}}?action=reset' \
      --header 'Content-Type: application/json' \
      Success Response::
        Response Status 200 OK
        Payload in body
      {
        "replication_id": "db1-rep-id2",
        "docs_read": 0,
        "docs_written": 0,
        "doc_write_failures": 0,
        "doc_write_conflict": 0,
        "status": "stopped",
        "rejected_by_remote": 0,
        "rejected_by_local": 0
      }

      Skipping TLS Certificate Verification

      Development and Testing Option ONLY

      This is an unsupported configuration option. It must not be used in a production environment. Its ongoing availability is not guaranteed.

      The configuration setting. database.this_db.unsupported.sgr_tls_skip_verify, can be used to skip the validation of TLS certificates, simplifying development and testing — see: Example 10 and the configuration item unsupported.sgr_tls_skip_verify.

      Example 10. Using sgr_tls_skip_verify
      {
        "databases": {
          "db1": {
            "server": "couchbase://localhost",
            "bucket": "db1",
            "username": "Administrator",
            "password": "password",
            "unsupported": {
              "sgr_tls_skip_verify": true
            },
            "replications": {
              "repl1": {
                "direction": "pushAndPull",
                "remote": "https://remotehost:4985/db1",
                "continuous": true
              }
            }
          }
        }
      }

      Handling Channel Access Revocation

      Users may lose access to channels for many reasons, including:

      • The User loses direct access to channel

      • The User is removed from a role

      • A role the user belongs to is revoked access to channel

      By default, documents are not auto purged on the active sync gateway even if the user on the passive sync gateway loses channel access.

      Note: that users are cluster-specific; userA in custer A is not the same entity as userA in cluster B.

      If required, you can override this behavior using the configurable option (enable_auto_purge-true).

      The behavior of the config flag is the reverse of what is done on Couchbase Lite.

      Using this option auto-purges documents on the active Sync Gateway that are no longer accessible, unless a document belongs to another of replicating user’s channels. This applies even if they are not actively replicating that channel.

      When enable_auto_purge-true=true, documents in revoked user channels are auto purged from Sync Gateway.

      This is consistent with Sync Gateway's handling of document access revocation using the purge-on-removal option

      Access Reassignment

      Where a user loses access to a channel and is then reassigned access to a channel, any previously auto-purged documents still assigned to any of the user’s channels are automatically pulled down by the active Sync Gateway.

      This will not impact active nodes that have turned off auto-purge behavior. Auto-purged documents removed from a user’s channels subsequent to the purge will not be synced again.

      If you want to control whether to sync previous auto purged versions of the document and do not want to pull down purged documents, you must remove the documents from all of the users channels to ensure they are not synced down again.

      Pull-only Replication

      The expected impact of the enablement of auto purge behavior when

      Scenario

      The replicating user of a pull-only replication is revoked channel access.
      ISGR is configured to run as admin user on active peer

      In ISGR, by default, access control policies are only enforced at remote cluster. The _sync function on an active node is by default run in context of admin user and as such there is no enforcement of access policies on the active side.

      System State Impact on Sync

      Active Sync Gateway (Local) (Running as admin user)

      Passive Sync Gateway (Remote)

      Expected behavior when enable_auto_purge is TRUE

      N/A

      User revoked access to channel

      Previously synced documents are auto purged on local

      Scenario

      The replicating user of a pull-only replication is revoked channel access.

      ISGR is configured to run as a non-admin user on active peer.

      Depends on availability of a new feature (3.0) wherein the active peer is also running as the replicating user.
      System State Impact on Sync

      Active Sync Gateway (Local) (Running as non-admin user user1)

      Passive Sync Gateway (Remote)

      Expected behavior when enable_auto_purge is TRUE

      User1 revoked access to channel

      User2 revoked access to channel

      Previously synced documents for User2 are auto purged on local

      User1 revoked access to channel

      Sync Function includes requireAccess (“channel”)

      User2 still has access to channel

      Config option has no impact.

      Previously synced documents for User2 remain on local

      Subsequent remote changes synced down are rejected by local

      User still has access to channel

      User revoked access to channel Sync Function access policy is a Noop

      Previously synced documents are auto purged on local