Printable View
Hi, not too familiar with linux from a sysadmin point of view. We have a box (RH 3.0) that regularly crashes around once a month, I can't see anything database/application wise but the sysadmin can't see anything their end either.
Anyone know any system logs I can look at etc?
Bazza
/var/log/messages
but if it crashes, might not be anythign there
Thanks davey, but you were right, nothing of value in there.
Hmm...back to the drawing board.
Bazza
Isn't it /var/adm/messages?
You could also try searching for core files which might be produced when the system crashes, and use strings or similar to try to discover what caused the problem:
find / -name core - to find any core files.
strings /path/to/core | pg - to read any readable text in the core file.
Hope this helps.
nope, redhat is definitely /var/log/messagesQuote:
Originally Posted by KenCunningham
Righto, ta!
Thanks Ken, I'll give that a whirl.
I have spotted something odd, Grid Control thinks OPEN_MAX has changed from 1024 to 65536, then to 1024 and then back to 65536 over a period of 3 weeks. The changes tie in with reboots/crashes but sysadmin don't know what causing this and didn't do anything. Yet if i do a getconf OPEN_MAX it returns 1024. It may just be a problem/bug with grid and yes I'm clutching at straws!
We run 3 databases, a couple of agents and grid on this box. We also see the same pattern on the DR server (and that also has the mysterious, ostensibly changing open_max value).
Bazza
Define crashes? we have servers that under high load 'freeze' and the ilo tools automatically reboot it after 5 mins. Slightly different to a crash
(all caused by unfixed redhat bugs)
crash in the sense you can no longer connect to the box or even do a simple ping. Everytime its happened, its been a case of powering the box on directly. The DR server was down for a matter of hours last time it happened on there.
No core dumps.
Redhat bugs...perhaps thats where I should look next.
Bazza