DCP Rollback in a loop, and oddly high seqEnd document

[AndrewLipscomb]

A sync gateway instance we are running reports DCP rollback message repeatedly, seemingly stuck in a loop

2022-09-11T22:58:43.571Z [WRN] c:bucket_name-SGI DCP RollbackEx request - rolling back DCP feed for: vbucketId: 762, rollbackSeq: 0. -- base.(*DCPCommon).rollbackEx() at dcp_common.go:187

I've had a look at the relevant document for that vbucket entry in Couchbase directly

  {
    "id": "_sync:dcp_ck:762",
    "bucket_name": {
      "failOverLog": [
        [
          0,
          0
        ]
      ],
      "seqEnd": 18446744073709552000,
      "seqStart": 0,
      "snapEnd": 0,
      "snapStart": 0
    }
  },

There are also other entries with similarly high seqEnd, 11 all up, with the same number

Obviously this isn't a lot of help without showing how this ended up with an oddly high number - if there is any other info that would be helpful to categorise this, let me know. That said we have been seeing some network instability recently, and this may have been due to an unexpected disconnection having some fun run-on-effects.

Sync Gateway version

3.0.3 - docker version

Operating system

Ubuntu 20.04 host, Couchbase Docker Image