k220 crash

[Umapathy S]

hi
I am running hp 10.2 on k220 class server. Yesterday night it crashed down mysteriously. I am trying to find out the problem. I fixed up blocks and restarted the machine. /etc/shutdownlog says that
13:10 Tue Feb 12, 2002. Reboot:
13:22 Tue Feb 12, 2002. Reboot:
There is nothing in /var/adm/crash. I saw something in /var/adm/syslog/OLDSyslog.log as
Feb 12 12:50:42 hpk220 syslog: getty: cannot open "tty0p4". errno: 6
Feb 12 12:50:46 hpk220 syslog: getty: cannot open "tty0p5". errno: 6
Feb 12 12:50:56 hpk220 syslog: getty: cannot open "tty0p3". errno: 6

this goes from tty0p3-p5. What are the reasons. Do I need to check more.

thanks in advance

cheers
Umapathy

[Patrick Wallek]

Have you checked /var/adm/crash to see if there are any crash dump files? If so, have you sent them off to HP to have them analyzed?

Your /etc/shutdownlog shows a reboot at 13:20 or so yesterday. What time did the crash occurr? If there is nothing at the time of the shutdown, I'd susepct a power supply or some other loss of power.

The messages from your syslog appear to be related to a terminal not being opened by a getty process. I doubt that those would cause a system shutdown.

[Umapathy S]

There was nothing in the /var/adm/crash directory, and I am sure that there was no power failure also

cheers
Umapathy S

[Michael Tully]

Hi,

Have a look in

/var/adm/syslog/OLDsyslog.log
/etc/rc.log (look for the crashdump,
it may indicate that it is not
configured correctly.)
/var/tombstones

The getty messages indicate that the
system can't start 'getty' processes
for those devices which look like modems.

-Michael

[Sanjay_6]

Hi Umapathy,

Chek your MUX card ports and its cable connection. Error no 6 means the sytsem is unable to open the port and hence cannot run getty on the ports mentioned. If you check you /etc/inittab file, you'll find getty running for these ports. But since the system is unable to access these ports, it is giving error no 6. If required replace the MDP/MUX Breakout box. If that doesn't help, replace the MUX CARD.

Hope this helps.

Regds

[Umapathy S]

yeah it says

Save system core image if needed
Output from "/sbin/rc1.d/S440savecore start":
----------------------------
EXIT CODE: 2 - savecore found no core dump to save
"/sbin/rc1.d/S440savecore start" SKIPPED

unfortunately when I did fsck and restated the machine, one log volume failed and I had to do fsck again. So my rc.log.old was rewritten twice. I didnt find any crash dump info there

[Umapathy S]

pardon me for my ignorance, what and why these MUX card ports being used for. If it is not needed for me then I can comment them out in the inittab file

cheers
Umapathy

[K.Vijayaragavan.]

MUX card ports are being used for connecting Dump terminals where users can login and for connecting external modems and serial devices.

If nothing is connected to these MUX port then you can comment those related entries in inittab file.
note:
tty0p4
tty0p5
tty0p3 all are representing the terminals that can be connected to mux port numbers 4,5,3.

-Vijay

[Michael Tully]

Hi,

You can comment out the entries in the /etc/inittab file. You also need to run the command

# init q

after making the changes.

It doesn't look as though we will find the exact problem at this stage. I cannot see why a MUX board would crash a system.

-Michael

[Printaporn_1]

Hi,

you say system crash in the night , then shutdownlog and syslog ,they got time stamp at afternoon. Does it can related ?
fom #who -b what time the system last boot ?
if who -b output > last time stamp in /etc/shutdownlog and
no panic and no HPMC/LPMC it can be power failure or somebody do TOC.

[Umapathy S]

yeah I found this just now also and I was wondering how it shall be, because I was working and I am sure I didnt do a reboot of the machine during the times that /etc/shutdownlog reports (ie midday between 1 and 2 pm).

date command returns proper time as of now
Wed Feb 13 13:28:46 IST 2002

Is there any problem with logger daemons.

If I suspect it as power failure then the UPS attached to my machine can be started manually only.

Is there any other way to check

cheers
Umapathy

[melvyn burnard]

When you say crashed, define the term crashed slightly better if you can.
was there a panic message on the console,?
Did the system just die?
was anyone working on it or near it at the time?

The log information shows there was no panic logged, and if there is no crashdumo, it points to one of the following:

1) power failure
2) hardware failure such as system board.
3) someone issued reboot -q command

[Umapathy S]

I was remotely working on the machine when the crash occured. my terminals were not responding for a considerable amount of time. When I tried to ping the machine, it was dead. So I ran to the server room. I am sure no one was in server room and I put on the console and saw the machine was in single user mode waiting to fix the disks.

I ran fsck, then rebooted the machine. Again there was an error with one logical volume. I fixed that with fsck again and rebooted, the machine came to life. Then I tried to telnet the machine, it said connection refused. I checked up for inetd, and not seeing that issued /sbin/init.d/inetd start.

Then the rest of the story, I was trying to figure why the machine crashed.... /etc/shutdownlog not showing me the correct time ..

cheers
Umapathy

[K.Vijayaragavan.]

Hi.

I think your server is still in single user mode (check by giving the command "who -r" and may be you have to make it to goto runlevel 3 (multiuser mode with network enabled) by giving the command "init 3" or "init 2" (multi user mode w/o network support.

-Vijay.

[K.Vijayaragavan.]

One more point,

As you were mentioning that you were working from the remote terminal which is in different room, you should not comment out the tty0p3,tty0p4,tty0p5 ,etc. from the /etc/inittab file.

If you comment it out then you may not be able to login to your system from remote terminals or PCs.

-Vijay

[Umapathy S]

Thanks all for your replies. Its my fault that I didnt prepare my system for a crash analysis. I will raise this again with details if any crash occurs in the future.

Thanks again for the time and response

cheers
Umapathy

[Sanjay_6]

Hi Umapathy,

Instead of commenting the mux entries in /etc/inittab, i would suggest you change the respawn entry to off for the mux ports for which the getty is running and which you don't want to use. Once you have done that, do a "init q" for the change to take effect. This will stop the getty for the ports to run by the system.

Hope this helps.

Regds