Server crashed - no crash files created

[Ra Jose]

I have Itanium server that crashed. I know as my xterm window to it went away. I have savecrash and crashconf configured. When the system came back up, no files were created in /var/adm/crash. I checked, SL event logs, old syslog, no trace of what happened. Any other ways to find out from system, why system crashed. Thank you for your replies.
Rajos

[Rick Garland]

Look in the /etc/shutdownlog file. What is the last line (or near the last line) saying?

Did system PANIC? Was it rebooted by someone else? What message is shown there?

Also look in /var/adm/tombstones. Is the a ts## file that has a recent date/time stamp? (This file could be named ts99). You could have HP look at the tombstone to see if there are any clues.

Look at the /etc/rc.log file. Any mention in there as to the crash files? Maybe the system tries to save them but something is not configured correctly. This will be a good place to see.

[A. Clay Stephenson]

Loss of a connection to an X session is no certain indication of a system crash. What does "who -b" report. There are cases where problems are so severe that no logging of any kind can be done but the very first thing to determine is if indeed you have a real crash. Did you check /var/adm/shutdownlog. The system may have been told to shutdown or someone could have simply issued a "reboot" command.

[Ra Jose]

The last lines in /etc/shutdownlog shows
16:53 Fri Jan 5, 2007. Reboot:
16:10 Fri Jan 19, 2007. Reboot:
The server crashed yesterday (Feb 21). No ts## files are there in /var/adm/ and also /etc/rc.log.old shows nothing about this. The crashconf and savecrash are correctly configured. Wondering how the ts## and /var/adm/crash files are not created. is there any daemon checked that does this.
Rajos

[Rick Garland]

OK, so the /etc/shutdownlog does not show any recent downtime.

What does your uptime value say?

No tombstone files? Are we sure the system crashed?

[Ra Jose]

yes Rick. Here is what happened. I had xterm session open to the server. It suddenly disappeared. So I tried to open another xterm session, but could not. Tried pinging, could not. So logged into the console. The system halted at the PDC level. It tried to execute "boot vmunix -lq" but exited saying "could not boot in maintenance mode" and stopped at this prompt. I had to issue "boot vmunix" to get it booted. Then I started troubleshooting what happened. So 2 questions. One, why it crashed. second, why it did not boot with "boot vmunix -lq" low quorum option. I know PA systems boot with this.
Rajos

[Rick Garland]

This is a call to HP support.

Let us know what happened.

[Patrick Wallek · Accepted solution]

First -- If there is nothing indicated in any logs about a panic, then it may be possible that you had a power problem of some sort. If power was interrupted the power supply (or power supplied) in the machine had a problem, then the machine would go down and nothing would be written to the logs.

If it was a temporary outage, then it may have come back on within a few seconds and the box came back up.

Second -- As for the '-lq' not working, that is usually an option to 'hpux' and not vmunix. My normal procedure when booting is to do a 'bo pri' or 'bo alt' and say 'YES' to Interact with IPL. Then at the IPL prompt, do an 'hpux -lq' to start in low quorum mode.

[Ra Jose]

Patrick. I found out what the problem was. Patrick's guess is right. There were power folks working in the datacenter and apparently, they were trying to plug in new servers for setup and plugged out our test server. Thank you all for your input.