topic Re: Unknown reason for system hang in Operating System - Linux

Unknown reason for system hang

Paul Thomson_2 — Fri, 21 Jul 2006 09:44:17 GMT

Hi

One of my servers running RH AS 3 Update 3
Kernel 2.4.21-20

The system was unable to be accessed via ssh or via ilo, once the password was entered the system would hang and never proceed past this point.

All I have to go on is the log /var/log/maillog displays messages such as
Jul 16 21:10:23 esmadl01 sendmail[1903]: rejecting connections on daemon MTA: load average: 48
Jul 16 21:11:08 esmadl01 last message repeated 3 times

This is repeated for days until the load reaches 1015 !!

I have no real idea what caused this, but we do collect some sar data this stopped at 19.30 on the same date (Jul 16)

I have attache this file sar.txt

The only thing strange is dentunusd and pgins goes high, Any ideas why the system would do this ?

Any thought as a little stuck
dmesg / messages shows nothing :-(

Thanks

Re: Unknown reason for system hang

Paul Thomson_2 — Fri, 21 Jul 2006 09:50:23 GMT

Also, this server runs veritas volume manager, and oracle 9.2

This server uses no external storage, only the internal disk cciss0: HP Smart Array 5i Controller.

Re: Unknown reason for system hang

Alexander Finkel — Fri, 21 Jul 2006 13:03:54 GMT

Paul,

Take a look at this:

https://bugzilla.redhat.com/bugzilla/long_list.cgi?buglist=117400

Looks like there might be a race condition with Veritas and certain versions of the 2.4 kernel where dentry_stat is updated to a negative number.

- Alex

Re: Unknown reason for system hang

George Liu_4 — Tue, 25 Jul 2006 13:14:25 GMT

The sar output shows nothing interesting. If the system is able to work after power cycle, is all file system, such as size, missing links, etc. normal?
I did see one Dell 6850 lost filesystem when running, but it is fine after it is rebooted.
Check to see if there is any kernel update available.