Open zeelax opened 9 years ago
Oh, forgot to mention some versions Elasticsearch 1.3.5 (tried all versions from 1.1.1) river-mongodb 2.0.2 (tried all versions from 1.7.3)
@zeelax please do let us know if you find anything. neither richard nor myself have time to look into issues unfortunately, but if you can find the problem, we're happy to help fix it, review pull requests, etc.
Having the same issue, caused a lot of hassle with ES having incorrect data as mongo updates weren't going through. Removed ("secondary_read_preference": true) option for now.
I have a shrarded mogno cluster setup. Three shard each contains of three node replicaset.
Everything is fine while tailing oplog from primaries. However, tailing from secondaries ("secondary_read_preference": true) is not so fine. Some updates are missing.
In river logs I can see right timestamp with proper ObjectId from mongo's oplog, but with data from previous oplog entry for this ObjectId.
I suspect that replication lag have some influence on how mongodb driver getting oplog from replicaset. Doing further investingation.