topic Re: My Linux is Dying in Operating System - Linux

My Linux is Dying

Maxim Rozin — Sun, 11 Jul 2004 04:59:06 GMT

Hello people,

There are many problems with my Red Hat AS 3 Server (runs on hp Proliant DL580):
First, it does not accept new connections (telnet, ssh). However, existing connections are working. Web connections work either.
Second, when I run xinetd -c, the proccess (xinetd -c) remains in the system (PPID=1).
Third, /var/log/messages is empty.
Another problem is that the load average is about 140 (cpus are 99% idle).
If that's not enough, crontab is not working too.

It happened also a week ago, and reboot solved the problems.

Anybody has a idea what is wrong here?

Re: My Linux is Dying

Vitaly Karasik_1 — Sun, 11 Jul 2004 06:57:36 GMT

please send us

1) "df" output
2) top [top -b] output
3) "free" & "vmstat" output

Re: My Linux is Dying

Maxim Rozin — Sun, 11 Jul 2004 07:42:39 GMT

Can u please tell me what exactly to look for (the server is not connected to the internet).

There are two more things we noticed about the server: when we kill a process, it becomes defunct;
And there are about 20 CROND processes in the system.

Thanx, Maxim.

Re: My Linux is Dying

Steven E. Protter — Sun, 11 Jul 2004 14:38:25 GMT

You need to boot the box.

Then you need to look at its cron schedule and see what its running and if there is enough time between runs on repetitive processes.

You need to look for full filesystms on the df -h especially /var because you can't run your system without /var

You more than likely have a cron job calling itself over and over again.

or worse, you have an authorized or unauthorized user doing the same thing.

SEP

Re: My Linux is Dying

Vitaly Karasik_1 — Mon, 12 Jul 2004 00:56:54 GMT

Maxim, you can transfer files via floppy disk; such logfiles will help us to understand the problem.

And regarding cron - there are two ways to deal with "too many crons":
- just take off cron : chkconfig crond off; chkconfig anacron off; service crond stop

- check crontabs of all users and delete problematic part.