If you add to the Chef Server’s runlist chef::server running the latest cookbooks from github.com/opscode/cookbooks.git it will do a compaction if your chef.couch is > 100megs on each client run; which will greatly help in avoiding said situation.
Additionally there is a bug in apache’s system http://issues.apache.org/jira/browse/COUCHDB-275 … basically you would need to update erlang in order to fix it… or you run couchdb 0.8, and that with the bugged erlang can cause 20gb couch files that won’t compact.
But basically if you use the chef::server recipe it will help ensure your couchdb won’t grow too big
On Dec 30, 2009, at 2:30 PM, Matt Matson wrote:
Thanks for the feedback. BTW, didn’t mean to slam couchdb logging… just that I’m not familiar with reading them yet.
So, I did add a recipe to compact couchdb. After you metnioned that, I noticed it was growing excessively large. File size went from ~2GB to 2MB after compacting!
I’ll continue to monitor the couch instance and report back if it continues to die. Thanks for the tips!
On Mon, Dec 28, 2009 at 3:05 PM, Adam Jacob email@example.com wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Has anyone noticed couchdb randomly dying? The logs aren’t helpful at all. I’ll continue to look into it… just thought someone may have encountered this problem before.
BTW, I’m running on CentOS 5.3…
[me@chef001 ~]# rpm -qa | grep chef
[me@chef001 ~]# couchdb -V
couchdb - Apache CouchDB 0.9.1
Can you confirm that you are compacting couch?
Adam Jacob, CTO
T: (206) 508-7449 E: firstname.lastname@example.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----END PGP SIGNATURE-----