Hi @nickva, thanks for following up. To answer your questions: \- Cluster don't change (always the same 3 nodes); nodes names (or IPs) don't change (more generally, we don't touch `vm.args` nor `/_membership` nor `/_nodes` nor `:5986` at all. \- No other tool or process that touch `/var/lib/couchdb` (except us when debugging, but only rarely and in read-only: `du`, `df`, `scp`...) \- Databases are not rapidly created (around 250 new DBs per day), and never deleted.
**Here is more info on the shard-sync issue:** We've set up a new cluster, and replicated from the (dying) old one. Three nodes, `q=8`, `n=3`. For a database chosen at random, there are **32** `_local/shard-sync` docs on first node ([download them all here](https://github.com/apache/couchdb/files/2423565/local_docs.json.txt)), among which some look redundant (same `source_node` and `target_node`, but different UUIDs). Example: ```json { "_id": "_local/shard-sync-1MHFZMU82CR6nTRw6IB5ig-MAqWsRU1KdFyVAmQyHUkqg", "seq": 68, "target_uuid": "147ecd8ddf1ba86c3e8d7117587141b1", "history": { "[email protected]": [ { "target_node": "[email protected]", "target_uuid": "147ecd8ddf1ba86c3e8d7117587141b1", "target_seq": 70, "source_node": "[email protected]", "source_uuid": "4ceab4f79e99ef3929a3f3e7bb9ea8f5", "source_seq": 68, "timestamp": "2018-09-27T07:34:01.554723Z" } ] } }, { "_id": "_local/shard-sync-6Fr1fwkrRyD7NMmRmXXCmQ-E-HElyKWVluvtMH9BuGyNA", "seq": 73, "target_uuid": "2d5e533605c004d673c338e796130b9d", "history": { "[email protected]": [ { "target_node": "[email protected]", "target_uuid": "2d5e533605c004d673c338e796130b9d", "target_seq": 75, "source_node": "[email protected]", "source_uuid": "7c44291538e9f88de564b13642122f12", "source_seq": 73, "timestamp": "2018-09-27T07:34:00.978796Z" } ] } } ``` Is it normal that there are 32 of them? Is it normal that some look very similar? [ Full content available at: https://github.com/apache/couchdb/issues/1621 ] This message was relayed via gitbox.apache.org for [email protected]
