@adrienverge
Thanks for your report. A lot of local documents could increase the file size
like that.
Do you know which type you have more of _local/<id> or _local/shard-sync-<id>?
If it is the first kind (_local/<id>), then those are the result of using the
database as a replication endpoint. CouchDB writes _local checkpoint documents
on the source and target, so next time it restarts the replication between
source and target it will resume from changes feed from the checkpointed value
and not from the very beginning.
However that means that each {source, target} pair (along with a few other
thing like if there a filter) generates a _local document. If say source is
used for a public demo, then there could be a very large number of {source,
target} pairs generated and thus a large number of _local/<id> documents
present.
There is an option to avoid this
http://docs.couchdb.org/en/stable/config/replicator.html#replicator/use_checkpoints
and it could help in these cases. Read additional info and the warning
regarding using that option, but it sounds like it might help your case.
If you see more of the second kind (_local/shard-sync-<id>), then it is hard to
say without more debugging why that's happening.
[ Full content available at: https://github.com/apache/couchdb/issues/1621 ]
This message was relayed via gitbox.apache.org for [email protected]
