L1000 server hungs often.

[Andrikopoulos Andreas]

i own an L1000 server with 3GB RAM, CPU 1X360MHz with HP-UX 11.11 which runs an application for network maintenance. The problem is that this server hungs really often. It answers to ping but i cannot connect to it via telnet. What i usually do is to log on to the GSP mode and do a restart. After the restart the sever behaves normally and i cannot find a problem on the log files. Two weeks ago i changed the cpu to a 440MHz but with no success. The strange thing is that i have another server with the exact configuration like the 'faulty' one which causes no problem.
What do you think i should do to investigate and solve the problem.

[Peter Godron]

Hi,
please compare your kernel values.
You may be running out of resources (open files,processes).
Go into SAM and compare the tunable parameters.
Try running the machine without the application to see whether it is a machine or application problem. Is there nothing worthwhile in the log files at all?

[JASH_2]

Hi,

Another way of checking your kernel files would be to copy the /stand/system file on one of the servers to say /stand/sysytem.servername then ftp it to the other server and then:-

diff /stand/system /stand/system.servername

Other wise Peter's idea is good.

Does it actually error when you are trying to telnet, or just sit there hanging?
Regards,

JASH

[Torsten.]

Have a look into the logs and search for error messages.

(/var/adm/syslog/syslog.log and GSP>sl)

[Andrikopoulos Andreas]

Thank you all for your replies. I will try to compare the kernel values through SAM and make all the suggestions you made. To answer to your question JASH when i try to connect through telnet it doesnt return an error message it just hungs. I checked the error logs and there is nothing suspicious there.

[Patrick Wallek]

Can you log in via a serial console? If so, the first thing I wold check is the inetd daemon to see if it is running.

[Florian Heigl (new acc)]

You could also just pull a crash dump instead of resetting.
This dump can later be analyzed either by HP or You.

instead of typig "RS" type "TC" at the GSP prompt when the system hangs for the next time.

You'll find information in /var/tombstones/ts99 and /var/adm/crash/crash.0

Review it and forward it to HP.

[DCE]

If you have glance installed, look at the system tables (lower case t) to see if you are running out of any resources.

Also, when the system hangs - shut it down immediately - and try pinging it again. It is a long shot, but some other system may be trying to use the same IP

[Jeff Schussele]

Hi,

I would suspect a runaway process.
Install MeasureWare on it & collect data to confirm that.
Of course you'll need a PerfView server to reap it.

My $0.02,
Jeff

[Andrikopoulos Andreas]

thank you all for your answers. I will give them a try and i will report you back on what i found. thank you very much