I have a mystery that I've been working all day and cannot resolve. My VMPro server is working fine 90% of the time, but once in awhile the machine hangs. About every 10th time I get pop3 mail, the process sits for 30 to 40 seconds and then completes. When I log in from console, it usually works fine, but every now and then it hangs for a similar time period. If I ssh in, similar results, and sometimes while ssh'ed in if I try to run something it hangs for awhile, then resumes.
The CPU & memory load on this machine ( 8 cores, 6GB memory, intel-64) is very low. Over 4GB free and CPU stays around a few percent. I've turned off clam and changed spamassassin to use smapd just in case it is load related. It may be a little better now, but that may be my imagination. Spamd occasionally kicks one of the cores up to 35%, but mostly the machine is just idling.
None of the logs in var/log are reporting any problems that I can correlate with the pauses.
I've poked around for VM setting that can be limiting concurrent processes or connections, but haven't found anything yet that would have system wide effect. I'm not as familiar with CentOS as I am with Unix, but I've started tearing into sysctl settings to see if there is soem CentOS limit not tuned for a server.
Anyone have any advice?
Steve
[edit] as soon as I wrote this I remembered the dovecot settings. I changed the max login processes from 128 to 256 and it appears to be working better... but it will take awhile to make sure.
Howdy,
You may also want to take a peek at some of the logs -- /var/log/messages and /var/log/maillog, to see if any errors are being logged around the time you're having trouble.
You may also want to view the "dmesg" output to see if the kernel is logging any errors there.
-Eric