[LRUG] Sidekiq / Heroku / Database Pool size.

Tom Armitage tom at infovore.org
Sun Jul 21 11:01:45 PDT 2013


I've got an app in production on Heroku. I'm using the "Crane" db,
primarily for its cache.

As far as I can tell, Unicorn is set up to have 3 workers per dyno, and 10
connections in the pool. I've set that 10 as a config variable, and my
Procfile is telling sidekiq to use a DB_POOL of 10 thus:

worker: env DB_POOL=$DB_POOL bundle exec sidekiq -q default

and concurrency in config/sidekiq.yml is also set to 10.

I'm *still* geting Sidekiq chucking ActiveRecord::ConnectionTimoutErrors
like so:

could not obtain a database connection within 5 seconds (waited 5.000126085
seconds). The max pool size is currently 5; consider increasing it.

I have no idea what's happening - it's clearly only Sidekiq that's not got
enough DB concurrency, but I've done everything recommended to alter it.

Any thoughts? This is happening frustratingly regularly now, and causing
issues in production.

t.

-- 
Tom Armitage
http://infovore.org
07813 060578
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lrug.org/pipermail/chat-lrug.org/attachments/20130721/02c634d0/attachment-0003.html>


More information about the Chat mailing list