Server Problem

[Andreas Voss]

Hi,

last day i had a problem with my server:
K220
2 x CPU
2 x 4GB mirrored (vg00)
HASS with 4 x 9GB mirrored
DLT7000 on dedicated bus
HP-UX 10.20
OmniBack II 3.10
SAMBA 2.0.7

The system gaves no responds to NFS and SAMBA clients.
Frontpanel said F72F (think F7 is heavy load)
I logged in via telnet.
Looked at top but 80% idle on both processors
Looked at ps -ef an saw 79! /etc/opt/resmon/lbin/registrar processes.
Looked in /var/adm/syslog/syslog.log: nothing critical.
Looked in /etc/opt/resmon/log/client.log:
-------------------Start Event--------------------
Event 132 occurred at Wed Jul 26 21:03:47.139024 2000
Process ID: 1715 (/etc/opt/resmon/lbin/p_client) Log Level: Error
do_version_handshake: Failed to receive a version reply
-------------------End Event----------------------

-------------------Start Event--------------------
Event 2505 occurred at Wed Jul 26 21:03:47.179300 2000
Process ID: 1715 (/etc/opt/resmon/lbin/p_client) Log Level: Error
update_monitors: socket open failed
-------------------End Event----------------------
This messages repeated very often

Finally i rebooted the system and now it works fine.
I also had a look at stm tools but no event found.

Anybody any ideas what happened ?

Thanks

[Patrick Wessel]

Andreas,

Are you running NIS on this box? In that case add the registrar service entry to the NIS Services map
-- ensure that:
ypcat servcies | grep reg
returns the correct service ( registrar 1712/udp # resource monitoring service),
-- Check the /etc/inetd.conf file to ensure the registrar entry is there: ( registrar stream tcp nowait root /etc/opt/resmon/lbin/registrar /etc/opt/resmon/lbin/registrar)
-- Have the inetd reread the file (inetd -c)
-- Rerun netstat -a | grep reg
-- Run /sbin/init.d/ems stop and then start

[Andreas Voss]

Hi,

i'm not using NIS.
/etc/inetd.conf is ok.
I put the ps -ef list at the attachment.
Could the problem have to do with the version of the OnlineDiag ?
swlist:
OnlineDiag: B10.20.15.10
OnlineDiag.EMS-Core: A.03.00.01

Thanks

[Patrick Wessel]

Andreas,
STM would have been my second guess. Did you ever update the Diagnostic on that box? In that case it might be possible that the file /etc/opt/resmon/dictionary/DEFAULT is in a bad shape. Is there a STATE_TYPE entry for each of the resources in this file?

[Andreas Voss]

Hi,

yes i've updated Diagnostics a time ago.
/etc/opt/resmon/dictionary/DEFAULT seems to be ok but it is dated from 20.08.1998.
It put it at the attachment

[Patrick Wessel]

The Copyright information is always 1998. But it shows " A.03.00 EMS DEFAULT Dictionary" and you wrote above that you EMS version is A.03.00.01 (which fits to the STM version)
Nevertheless I compared the file with a A03.10 version (we have no 3.00.01 installed) and found the following difference:
RESOURCE_NAME: /cluster
DESCRIPTION: "Cluster Resources"
STATE_TYPE: B

No idea if you really need it. But maybe you have another system with EMS A.03.00.01 running without problems. Do you? In that case you could compare/copy this file from the other server

[CHRIS_ANORUO]

Andreas,

One thing you have to crosscheck are the patches and their dependencies for this version of OnlineDiag.


Chris

[James R. Ferguson]

Andreas:

This may be a long shot. I recently updated Diagnostics on my D- and K- boxes running 10.20. I later (after some weeks) found that STM wouldn't initialize correctly. The problem was that I hadn't specified reinstall=true. HP informed me that for Diagnostics this is 'standard/preferred'. Have you tried reinstalling these?

...JRF...

[Andreas Voss]

Hi James,

I don't know if this option was used because a hp engineer had installed the software.
I think i look for the newest one and install it myself.

Thanks