Hello Is anyone successfully using auto process scaling on a high volume system?
I've been having a number of issues, including excessive system load, deadlocks on the UDP worker processes and segmentation faults, both once running as well as in the attendant (starter) process.
While I haven't yet been able to establish a conclusive reason, one of the changes that I implemented in the last week was to enable an auto_scaling_profile.
Unfortunately my init scripts weren't set up to collect core dumps, so I cannot establish with any certainty the exact point at which the past segfaults occurred. I have now corrected these so that, going forward, I should be able to obtain backtraces. But there mere fact that average system load has been ~33% less in the hour since restarting with auto scaling disabled versus the hour prior (despite traffic levels being similar) is, in itself, quite curious.
Thanks Greg _______________________________________________ Users mailing list Users@lists.opensips.org http://lists.opensips.org/cgi-bin/mailman/listinfo/users
