Site performance, Server errors, outages and tunings

[Like Ra]

I hope I've found the source of the "Server error"s. It was a hard procs per user limit in httpd.conf. Let's see if removing this limit helps.

[Like Ra]

The errors disappeared from error_log and the load jumped a bit up. Looks like apart from the overloaded disks (RAID 6 failure) the WHM limits was the problem #2.

[Like Ra]

There are two choices: either "Server error" or "Site down" (when overloaded). The latter occurs rarely, the former more often. Let's try the "if overloaded, it's overloaded, just restart apache" approach.

[Like Ra]

I'm pretty much happy with how it works. I noticed several mysql/php errors, but they happened because of a huge overload.

Also I adjusted my monitoring script.

[Like Ra]

You may have seen mysql errors. Found some issues with the data base. Trying to fix them.

[Like Ra]

Forgot to add - yesterday I discovered that VPS was completely down for unknown reason. I booted it up, but the admins were doing something in the system afterwards. No idea what and why.

[Like Ra]

Observed interesting mysql behaviour - when restarted it takes 500MB resident memory (and almost 3GB virtual), after running for 16 hours the RES is back to 17-20MB.

[Like Ra]

After the disaster yesterday (http://www.likera.com/blog/wp/archives/11068 ) everything looks fine and fast again.

[Like Ra]

Aaaaaannd two more outages!!! I reckon the hoster hates me, because I raise tickets and ping admins for every hiccup.

Limited the maximum number of processes again.

[Like Ra]

Removed a couple of modules from Apache and added buffered logging.

OTOH, I love performance tuning 😉