redhat 9 dies randomly

Bill McCormick wpmccormick at sbcglobal.net
Tue Aug 10 03:58:50 UTC 2004


kalin mintchev wrote:

>hi all..
>
>i posted this question a few months back here and the only answer i got
>was to update the kernel which i did but it didn't solve the problem..
>
>i have a redhat 9 machine that dies randomly. sometimes after a day,
>sometimes after 8 days. there isn't any indication of a problem in any of
>the logs.
>all the machine runs is qmail with spamd and some procmail and a bind name
>server. it also has a helix real media server but  that was added a few
>weeks ago. there is a web server with webmail interface - squirrelmail -
>too. that's all.
>non of the logs show anything. there is enough disk space. i did a memory
>test  which came out fine.
>
>now having said that i heed to mention that a few weeks ago the machine
>had a second drive on which the main swap partition was living. that drive
>died and i had to create a swap file under /var on the main disk. i
>thought this will solve the problem but apparently it's not.
>
>the machine is monitored 24 hours but sometimes people notice that the
>mail is down before the techs - just a minute or so after the machine
>dies. and rebooting it all the time is not solving the issue. of course i
>can rebuild the machine but it be nice to know what the problem might be..
>
>can somebody provide some advice of where to start looking for a problem.
>i'd appreciate it.
>
>also is there a way to find out what was the cpu load or memory usage at
>the time of the crash?
>
>thank you...
>  
>
I had this exact same problem a few months ago. Red Hat 9, qmail, SA,
procmail. I also had fetchmail going ass well as a dyndns client
(ddclient). I racked my brain for weeks I swear. I did all kinds of
methodical testing to try to narrow down what process was causing it.

I can't be for certain exactly what fixed it, but I recall adding an MX
record specifier in my ddclient.conf and then things were back to
normal. It kind of made scense in a way. I recall determining that this
problem only happened when qmail was running. One obvious simpton was
that SA would take for ever (like 30-45 sec) to process one e-mail. I
think it's also possible that running named (DNS) was also helpful, but
I can't be certain.

Hope that helps.


Bill





More information about the redhat-list mailing list