Python's threads do not span over multiple cores/CPU's, so you are correct in only seeing 100%. Maybe you could try a cluster solution and keep a dedicated node for the webserver? (http://bitbucket.org/galaxy/galaxy-central/wiki/Config/Cluster) Kanwei On Wed, Jan 20, 2010 at 3:24 PM, Assaf Gordon <gordon@cshl.edu> wrote:
Me again, my Galaxy server is still slow, and I'm desperately looking for possible solutions.
Any help, suggestion or idea (from anyone) will be highly appreciated.
The metadata is now set externally. The database is postgres, switching to MySql showed minor improvement (especially after reboot, where all cache was probably used by MySQL).
The python process goes to %40 while few jobs are running, and goes to 100% when a lot of jobs are running and a lot of users accessing galaxy (list their histories, run tools, workflows, etc.).
One strange thing that I'm seeing, is that the python process never goes above 100% (actually 106% but I suspect some 'top' summing inaccuracies). Enabling threads-view in top (pressing 'H') shows all the python threads, but their CPU usage *always* sums up to 100% - never higher.
I would assume that a true multi-threaded application can easily go above 100% - each thread should be able to reach 100% independently.
Could it be a problem in my python installation ? maybe it doesn't support threads correctly ? or is this a wild goose chase?
Thanks, -gordon _______________________________________________ galaxy-dev mailing list galaxy-dev@lists.bx.psu.edu http://lists.bx.psu.edu/listinfo/galaxy-dev