System Panic

[Rusty Sapper]

I've had a server crash today and the only real evidence is an entry in the /etc/shutdownlog which says:

Reboot after panic: callout table overflow

can anyone give me more information?

this is an N-class running 11.0

TIA


Rusty

[Trond Haugen]

Not having a dump I can only guess on what was the cause. I would suggest checking if PHNE_24181 is installed and increase the kernel parameter ncallout.
Then I would look into why there is no dump. Check 'crashconf -v', /etc/rc.config.d/ crashconf & savecrash and try running 'savecrash -r'. If you succeed in getting a dump I would recommend contacting your local response center.

Regards,
Trond

[Mark van Hassel]

Did you check /var/adm/syslog/OLDsyslog.log ? There should probably be some "table overflow" messages in it.

HtH,

Mark

[Tom Geudens]

Hi,
Patch PHKL_22330 might help (if you don't have it yet :-).

Regards (have been trying for a day to get this posted :-),
Tom Geudens

[Magdi KAMAL]

Hi Rusty,

You have to increase the value of your kernel parameter "ncallout" which is "The max number of pernding timeouts".

I don't know how much the value of this parameter is on your system, but you may still need to increase it by 30%.

Magdi

[Bill Hassell · Accepted solution]

ost likely, the size of kernel parameter ncallout is way too small. There is no way to determine the underlying cause (such as threaded application programs, MUX drivers, etc) without getting a crash dump analysis from the HP Response Center. Double otr triple the value for ncallout.

Make sure the patches are up to date (ie, Support Plus CDROM or download from ITRC) and if you are running lots of threaded programs, bump up ncallout to 4000 to 8000 and nkthread to a similar value.

[Ross Martin]

ncallout may need to be increased in the kernel.

NCALLOUT (16+NPROC) NPROC (20+8*MAXUSERS)
MAXUSERS 32

Any of the above can be modified

Adjusting it is trial & error.

Hope this helps.

Ross Martin
HP Repsonse Center
Check ITRC for further documents and helps.

[Sanjay_6]

Hi Rusty,

Try this link.

http://us-support3.external.hp.com/cki/bin/doc.pl/sid=94e87b400c53646357/screen=ckiDisplayDocument?docId=200000056242887

Hope this helps.

Regds

[U,.SivaKumar]

Hi,

The reason for the crash is number of pending timeouts on the system might have exceeded the maximum number of allowable pending timeouts on the system which is determined by kernel parameter ncallout .

Try increasing the value of this ncallout parameter. HP recommends to set ncallout to a value greater than the sum of nkthread + the number of I/O devices connected to the machine.

regards,

U.SivaKumar.