Tracking down a resource starvation issue

Question

My Ubuntu 9.04 server on Slicehost is periodically freezing, and not responding to ssh or console access. I believe it is some kind of resource starvation or swap death issue.

This has been happening every few days, and I need to figure out what is causing it. Can anyone recommend the simplest way to do this?

Someone recommended "sar", and I've installed it, but its not clear to me whether it can pinpoint a specific troublesome process, or whether it is running frequently enough to catch the likely culprit.

Answer 1

To graphically identify any particular issues (this can be easier than just looking at sar data), use munin.

Also, install the process accounting package and look at the output of sa to see what is consuming so much memory, cpu time, etc, if anything:

apt-get install acct
update-rc.d acct defaults
/etc/init.d/acct start

All the details and output of the sa command can be found in man sa.