HP-UX 11 Networking Setup

[Ray H. Gay]

I am new to HP-UX and have an old 11.0 system that has the following problems I need to fix:

Routinely on a daily basis, it stops accepting telnet connections. Existing ones stay up. The console terminal becomes unresponsive and we have to power it.

When it does come up, I have to start the main ethernet (lan0) interface manually.

Then the NIS complains "Unable to bind to NIS server using domain " and NIS client never comes up. (from rc2.d/S420nis.client start)

Is there a good recipe for configuring tcp/ip from scratch that tells me what files to touch?

Thanks.

rg

[Zygmunt Krawczyk]

Hi Ray,

you can configure networking by invoking a shell script:

/etc/set_parms

Some basic network parameters are stored in /etc/rc.config.d/netconf

Regards,
Zygmunt

[Andrew Rutter]

hi,

you could also reconfigure the network through SAM. This is easier if you dont have much working knowledge of command line hpux.

You can unconfigure lans and reconfigure them as you wish setting all you would need for basic networking.

Also which system is it as there are many patches for hpux and PDC firmware updates to fix such things.

just a thought before you wipe the config only to find the same problems start to appear.

Andy

[Ray H. Gay]

Thank you guys. I got the network cleaned up and NIS removed as it was not needed. Using SAM was the easiest. I'll keep this open while we check to see if the connection failures continue.

[Ray H. Gay]

As suspected, the locking up problem is still there. It's running Oracle and Weblogic servers and quits taking connections about mid-day after people start working. The console (serial terminal) took a couple of commands and then locked up on a df -k | more.

What are the best log files to look at?

I'm thinking it may be something gobbling up memory until Unix can't allocate any for a session. Disk space still seems OK. Swap is 2GB and memory is 2GB.

[Andrew Rutter]

hi, well oracles known to be mem/proc hungry.

the syslog.log is a good place to start then etc/rc.log, var/adm/cron/log. and look for anything untoward. Also look for any large files still knocking about that that are not needed, and what process are running. You do get rogue processes hanging around from time to time that may need killing.

You could also double check the memory/processors are okay in CSTM.

What system is it and how many processors does it have and allocated.

andy

[Ray H. Gay]

It's a 9000/800/L1000 with 1 proc. All the logs look OK and no resource issues visible, but I do see there have been no patches since this system was born, so I am looking at PHNE_26096 which addresses similar failures with telnetd.

[Jannik]

do you have any nfs mounted on the system. If NFS mount have lost the connection to the NFS server the df -k will hang.

Do not just patch the system, this may introduce new problems, trie to find issues connected to the problem you have.

[Andrew Rutter]

hi again, in my opinion it would be good practice to bring the system upto date with patches, not just the one you mentioned but any others from the supportplus bundle.
**Dont forget to do a full backup first though.**
Also check the pdc and gsp firmware versions, these may need upgrading aswell. Check the firmware patch details to check for anything like the problems your having.

It may be also worth starting a new post in the Hpux sys admin forum to get the best advice with the oracle/ux problems.

Andy

[Ray H. Gay]

Thanks. I am on to hardware issues now using stm. I will open a new one for those.

[Ray H. Gay]

Hardware problems showed up finally.