Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Already on GitHub? Sign in to your account

Frozen state when out of mem (maybe?) #136

Open
hofmeister opened this Issue Sep 8, 2011 · 3 comments

Comments

Projects
None yet
2 participants

Just had an issue which caused c-l to never continue the view changes updates. (no log output - have a fork that will output at which change number you're at when timing out)

It seemed to occur after updating a view definition, realizing that it was wrong - and rolling back to the previous view definition.

Restarting c-l and couchdb did nothing good.

Restarting the entire machine solved the issue - and lucene starting indexing from the correct change number once more.

I was able to use the index (using stale=ok) when the problem was occuring.

Will be happy to assist in debugging - this was wierd

Info:
CouchDb 1.0.2
Couchdb-Lucene - fork from June 06 (0.7)
OS: Ubuntu 10-something
MEM: 24 GB
Disk: Lots
CPU: 6-core intel

1 more note: It involved 2 views - both on seperate dbs - both dbs contain 10 and 20 mio docs of significant size.

The error did reoccur soon after the reboot - following a massive "too many files open" stream of exceptions... And the machine does have all the ulimits etc. set to above 50K ...

Contributor

stefankoegl commented Jan 24, 2013

I've noticed a similar issue; after a restart, c-l would just log

2013-01-24 16:51:27,649 INFO [Config] Index output goes to: /usr/local/couchdb-lucene/indexes
2013-01-24 16:51:27,681 INFO [Main] Accepting connections with SelectChannelConnector@localhost:5985
2013-01-24 16:51:46,967 INFO [mygpo] Indexing from update_seq 22091566

an then freeze. As there seems to be no quick solution (no replies to this thread, and non discovered myself), I'll try by recreating the index

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment