I found a possible problem: databse access: my jobs where accessing database.
i have 20 sidekiq worker.
I put a pool_size for the database of 22 (1 web worker, 1 rake console, 20 workers .. in my yml) and also 22 database connections in mongD. It solved the problem. What takes before 1 hours take less than afterwards 10 minutes.
mongo.yml:
staging:
adapter: mongodb
database: database_name
host: localhost
port: 27017
options:
pool_size: 22
$ cat /etc/mongodb.conf |grep max
maxConns=1000
It may not be the more efficient settings, but I didn't have enougt time to tune them!
Aucun commentaire:
Enregistrer un commentaire