topic NODE_TIMEOUT in Operating System - HP-UX

NODE_TIMEOUT

Lorayne — Tue, 29 Nov 2005 07:49:03 GMT

Hi

I am trying to run check script on two servers in cluster. This message is displayed

./check.txt
********************************
*** Script to check firmware ***
*** PDC etc. ***
** @(#) Type G 22/12/04 ***
********************************
Files will be created in : /tmp/roota16814
Available space : 318904 Kb
Logical volume : /dev/vg00/lvol4

This is a active Serviceguard cluster, checking NODE_TIMEOUT!
A value between 5 and 8 seconds is recommended for CPU intensive Apps.
Checking this using cmgetconf can take up to 10 minutes!
On process hang, check LVM patches, see wtec call 4000030515!
vgscan -p used in cmgetconf can be root cause!
No data from cmgetconf, looking for existing ascii-file(s)
No or too many files XXX.ascii found, cannot continue...
NODE_TIMEOUT variable not displayed during /usr/sbin/cmgetconf!
Possible a Serviceguard error situation!
Run /usr/sbin/cmgetconf which will display the error!
If help is needed, contact the nearest HP response center!

I need disable NODE_TIMEOUT, where is this variable?? Can you help me.. Thanks...

Re: NODE_TIMEOUT

Steven E. Protter — Tue, 29 Nov 2005 08:00:32 GMT

Shalom Lorayne,

Suggestion:
Check networking in general and see if there is a slowdown of any kind.
Check ping times and see if they are unreasonably high.
Check for SG patches and get them in.
State your version of both the OS, SG and latest quarterly update installed.
Check inetd.conf and make sure the latest changes for 11.16 SG are in if that applies to you.

I don't think changing the NODE_TIMEOUT to 10 minutes is going to do you any good. The underlying issue seems to be your Cluster is not working correctly and the root cause should be dealth with.

find /etc/cmcluster -exec grep -l NODE_TIMEOUT {} \;
# The find command will provide you a list of the configuration files containing the variable.

SEP

Re: NODE_TIMEOUT

Mark Nieuwboer — Tue, 29 Nov 2005 08:41:21 GMT

Hi,

Yes you can comment it out.
Just command the line out in the checkscript
when it checks for the mc-serviceguard.
vi the script and look for SERVICEGUARD.
If i'm correct you can put a 1 instead of a 0 in the beginning of the newest check file and mc-serviceguard wil not be checked.

grtz. Mark

Re: NODE_TIMEOUT

Kent Ostby — Tue, 29 Nov 2005 10:19:57 GMT

Check for NODE TIMEOUT can be accomplished in a couple of ways.

if you are running a new version of Serviceguard, you can run the command "cmscancl" which will create /tmp/scancl.out

grep -i timeout /tmp/scancl.out

will reveal the value.

Otherwise, check the /etc/cmcluster directory for your cluster ascii file and type:

grep -i timeout

Oz

Re: NODE_TIMEOUT

Lorayne — Wed, 30 Nov 2005 12:11:42 GMT

Hi,

Today I will do this. Thanks...