I have an elasticsearch cluster with five shards where the shards are hitting the ~2B document limit. Looking at the logs, a large fraction of those documents are deleted documents and I am hoping that I can remove some of those deleted documents via a merge to temporarily allow writes until I can do a full reindex into a larger cluster. I tried doing a "force_merge?only_expunge_deletes=true" query, which helped the situation somewhat - now most of the new index operations succeed, but many still fail with a 400 error due to the document limit. The deleted count initially dropped, but has remained the same for several hours. The _stats endpoint says that there are 5 merges currently happening, but that has not changed in several hours either.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.