Hi,

I am running fedora core and plesk on an apache2 server. My server hangs once each day. The following is a snapshot of my server-status file. For some reason, I have been getting a lot of processes that don't have any PID or ip address. Is this a problem? Please let me know

Srv PID Acc M CPU SS Req Conn Child Slot Client VHost Request
40-0 - 0/0/582 . 2.81 401 0 0.0 0.00 4.58 ::1 host.domain.com GET / HTTP/1.0
41-0 - 0/0/177 . 0.00 977 0 0.0 0.00 1.74 ::1 host.domain.com GET / HTTP/1.0
42-0 - 0/0/291 . 0.02 969 0 0.0 0.00 1.91 ::1 host.domain.com GET / HTTP/1.0
43-0 - 0/0/5 . 0.10 978 0 0.0 0.00 0.01 ::1 host.domain.com GET / HTTP/1.0
44-0 - 0/0/232 . 2.59 628 0 0.0 0.00 1.86 ::1 host.domain.com GET / HTTP/1.0
45-0 - 0/0/8 . 0.00 953 0 0.0 0.00 0.09 ::1 host.domain.com GET / HTTP/1.0
51-0 - 0/0/21 . 0.09 963 0 0.0 0.00 0.05 ::1 host.domain.com GET / HTTP/1.0
52-0 - 0/0/18 . 0.13 959 0 0.0 0.00 0.20 ::1 host.domain.com GET / HTTP/1.0
53-0 - 0/0/13 . 0.20 961 0 0.0 0.00 0.12 ::1 host.domain.com GET / HTTP/1.0
54-0 - 0/0/15 . 0.05 958 0 0.0 0.00 0.12 ::1 host.domain.com GET / HTTP/1.0
55-0 - 0/0/234 . 2.25 350 0 0.0 0.00 2.31 ::1 host.domain.com GET / HTTP/1.0
57-0 - 0/0/10 . 0.07 976 0 0.0 0.00 0.04 ::1 host.domain.com GET / HTTP/1.0
58-0 - 0/0/13 . 0.00 962 0 0.0 0.00 0.01 ::1 host.domain.com GET / HTTP/1.0
59-0 - 0/0/2 . 0.00 983 0 0.0 0.00 0.02 ::1 host.domain.com GET / HTTP/1.0

I know it's been over a month, but have you looked into doing some tweaking?

Things along the lines of:
- Long lost child / or orphaned process
- Server load / optimize processes
- Cleaning out orphaned packages
- Shutting down generally unneeded services (and keep them shut down), such as Tomcat.
- Cleaning out totally unnecessary things (such as Bluetooth)
- Log rotation
- Tweak httpd.conf
- Re-updating everything.

And so on...

Unfortunately I can't make much of that log, it's so much Martian to me but it might help further to upgrade your log level (inside httpd.conf) make it one or two levels tighter so less is logged. Possible values include: debug, info, notice, warn, error, crit, alert, emerg... These are in order of increasing severity, but on a production server even warn and error is usually too much. I normally set mine to either crit or alert, this way only the more important crap gets logged (and it's still plenty to sift through)... But this helps some.

Because if I'm not mistaken it's a load issue, your server gets X amount of traffic and X many processes come online...
Some never shut down properly and over time this builds and eventually takes the server with it, if traffic were heavier it might happen more frequently, less if lighter (or not, just at maintenance time or a certain cron process does it).

That would be my first step is tweak it on up, now don't delete anything before you know what it does but once a process that isn't needed is shut down that might be well enough, also I am sure you'll find stuff that can be safely removed so I would stick to that... I wouldn't upgrade too much but certain packages might call for it, do that, otherwise it's likely best with less.

Good luck