Community mailing list archives
Re: How to optimize general performanceby
Tbh Alex I do it slightly differently. Hard limit around 12gb, soft limit at 2gb. That way really big batch processes complete, but run of the mill requests don't take a lot of ram. That's with 25 workers and ram usage for python rarely exceed 20gb. But of course monitoring logs is really the only way to tune that to a specific workload. Although I'm wondering now if restarting workers clears cache and a higher soft setting may be better.
In terms of postgres I pretty much use the setting you suggest but I think it's largely ignored in recent versions except by the query planner. But it's what pgtune spits out anyway.
Odoo is clearly OLTP.Regarding memory: if there are 8 cores, plan for 8 workers, and configure a hard limit of 4GB per worker. This means 32GB for Odoo, say 40. You may assign ~64GB for postgresql.2016-02-05 22:34 GMT+01:00 Andreas Becker <email@example.com>:+ KittiWhat would be best to set:Web applicationOnline transaction processing systemData warehouseDesktopMixed type of applicationswhat is your experience?How many connections would be best?Total Memory - does it mean all memory of the system or if you run also other applications on the server which use. i.e. mariaDB or MySQL etc.only the memory you would like to give to postgres?+1 Kitti, very useful website.The 2 most simple, quick efficient things to tune are : - enable workers in the Odoo config, so you can use your multi-core CPU - configure Postgresql with PGTune, which will set sensible values for your hardware LS.