Hi @nickva,
I apologize for the misunderstanding, I meant 1000000 docs for the whole node
("in our servers"). I'm glad to hear 32-48 `_local/shard-sync` docs are normal.
We started using `use_checkpoints = false` and deleted all other local docs.
Unfortunately CouchDB disk usage still keeps growing (by multiple dozens of GB,
while the size after a fresh restoration is 5 GB) even on a cluster we don't
touch (no connection at all). We are still investigating it, but it may be
because:
- shard files (`db.1520354435.couch`) constantly keep growing (by ~kB), even if
there are no requests done on them,
- compaction doesn't seem to happen on all shard files. This could be related
to `min_file_size` or #1579 or #1383 or #1097.
So in the meantime I'll close this issue. Once again, thank you for your
detailed explanations that help us understand and debug this problem further!
PS: By the way, we're stuck on 2.1.2 (because 2.2 has #1550, will be fixed in
2.3), but 2.2 has many performance improvements. It would be great to have a
2.2.1 (with the fix) while waiting for 2.3! :)
[ Full content available at: https://github.com/apache/couchdb/issues/1621 ]
This message was relayed via gitbox.apache.org for [email protected]
