You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When multiple indexing jobs are running I noticed critical indexing problem.
I run search query ?q=*:*&limit=0 to reindex search index on newly replicated database.
Server starts indexing, response is returned but index is not created.
those are steps that shows problem:
run search query ?q=*:*&limit=0
wait for completion ~1 minute
check _design/DESIGNDOC_NAME/_search_info/INDEX_NAME endpoint - pending_seq is not the same as DB update_seq
again run search query ?q=*:*&limit=0 and i have to wait for index to recalculate
wait for completion ~1 minute
check _design/DESIGNDOC_NAME/_search_info/INDEX_NAME endpoint - pending_seq is not the same as DB update_seq
...
I stuck on this problem when replicating multiple DB's to new environment and tries to reindex all indexes in those DB's. It was about 50 db's with 1k - 1000k documents inside and from 10 - 150 indexes.
I wrote a program to reindex databases after replication, this program just called search queries on indexes 10 indexed at once. This is when those strange behavior happens.
In couchdb logs i have multiple info messages index xxx closed with reason lru, maybe this is relevant.
Clouseau is running on windows 2012R2 server.
The text was updated successfully, but these errors were encountered:
I think I found reason, in my refresh indexes tool I had a loop to refresh index info, this loop was quering _design/DESIGNDOC_NAME/_search_info/INDEX_NAME endpoints for all indexes.
It looks like querying _design/DESIGNDOC_NAME/_search_info/INDEX_NAME endpoint refreshes index in LRU algorithm and LRU removes index that was currently indexed -> progress was not commited...
When multiple indexing jobs are running I noticed critical indexing problem.
I run search query ?q=*:*&limit=0 to reindex search index on newly replicated database.
Server starts indexing, response is returned but index is not created.
those are steps that shows problem:
I stuck on this problem when replicating multiple DB's to new environment and tries to reindex all indexes in those DB's. It was about 50 db's with 1k - 1000k documents inside and from 10 - 150 indexes.
I wrote a program to reindex databases after replication, this program just called search queries on indexes 10 indexed at once. This is when those strange behavior happens.
In couchdb logs i have multiple info messages index xxx closed with reason lru, maybe this is relevant.
Clouseau is running on windows 2012R2 server.
The text was updated successfully, but these errors were encountered: