hp-ux 11.0 problem

[srinivas_34]

Hi,
i have hp-ux 11.0 on HP 9000/K360.The problem was it stops communication but the mechine is up and the console connetcted to that mechine also stuck and the error message is saying " possible network disconnection on lan0 and asking to check the network connection"
we checked the network connectivity and the network is fine.
SO since the mechine is not communicating eaither to check the whats the problem ???
so we removed the power cable and put back.
then the error message on the littile display at the server is as follows
" INIY C440
PRI CNSL CPU01 "
it stuck with that and its not going further.
can anybody advice me with some posssible solution.
thanks in advance for your valuable suggestions.

thanks

[Helen French]

1) First, pulling the power cable from an online system is not advisable. It will cause severe damage to your system and file systems. Normally, a network cable or other communication issue does not require a system reboot.

2) Now check your system for any hardware errors. Make sure that you cables are connected properly (network, console cable etc). Check the display pannel on the server and see if it gives any error codes.

[Steven E. Protter]

With regards to the network situation.

If there is even a short disruption of network service on the swtich or hub your HP box is plugged into, the message will appear, probably in /var/adm/syslog/syslog.log

If there has been no disruption, you may have a hardware issue with the card.

Some steps to help figure it out.

lanscan

lanadmin -x #

replace the # with the number from the lanscan display.

You should see something that indicates the speed and duplex state of your NIC card. If its not what you expect, its a problem.

NIC cards of that era should max out at 100 BaseT Full Duplex. They should be set manually, not autonegotiate. You need to check with network administration what the speed is on your hub or switch. It needs also to be set manually, not autonegotiate.

Once that issue is checked, you need to run some simple ping tests.

from another unux/linux box.
ping hostname

From linux you can test harder with ping -f hostname

You want to watch ping times and see if they rise or a lot of them get dropped. If that is the case, you have a bad NIC card and you need to arrange replacement.

SEP

[Kent Ostby]

C440 means that the system cannot talk to the console.

Here is what you need to do:

Power off the CONSOLE .

Power the console back up while holding down the "D" key.

Count to about 8.

Release the "D" key.

Hit return when you see the message about "default configs used".

The system should start to boot at that time.

Best regards,

Kent M. Ostby

[srinivas_34]

Hi WILSON,
Since we are not able to see any sort of problem with the server (Even system console is also stuck ) we happened to have only one option of removing power cable and put it back.
As for as network card/port is concerned its working fine.
When we shutdown and restart the error is showing on little display screen as "
INIT C440
PRI CNSL CPU01 "
PLEASE ADVICE ME FOR ANY SUGGESTIONS.
THANKS

[Patrick Wallek]

With the network message you were getting it means the lan is disconnecting for some reason. You should check the LAN card (it may be going bad), the network cable(s), the swith / router port.

[Helen French]

The error code C440 is normally a console connection issue. Check if you have "remote" mode enabled on the console. You can also hold the "D" key (for about 15 seconds) when you reset the console and then see if it works. Also, check the cable connections and terminal.

[srinivas_34]

Hi Kent,
Thanks for your reply.
Now my server is up and running and i tried to look the log files syslog but it does not give any problem relating to h/w .The log file is quite normal without any errors.
so now i am trying to find out why it stops communication from yesterday evening 4:20PM to till the mechine powerd off around 1100AM
is there any suggestions to traceout the problem.
thanks in advance for your advice.

[Kent Ostby]

You would need to check the OLDsyslog.log in /var/adm/syslog.

If there is nothing there, then you probably need to check the nettl.log file with the netfmt command.

Best regards,

Kent M. Ostby

[srinivas_34]

Hi KENT,
thanks for your speedy reply.
i checked the old logfile only and the last entry on the log file is at 4:20PM yesterday with normal reguler system logging messages with no problem relating

[srinivas_34]

Hi kent,
i just checked the nettl.log file and the last entry on that file is longback message relating to LAN/900 NETWORKING and the time stamp on the nettl.log file is not recent.
can u advice me what may be wrong that makes my mechine stop communication.

[Kent Ostby]

srinvas --

I am not sure.

In general, when your machine wont respond, you would need to do a "TC" or Transfer of Control to generate a hang dump.

To do that on a K-series, you would put the key in service and do cntl-B then TC to do a transfer of control.

NOTE: This will create a core dump AND reboot your system.

ALSO NOTE: Its only useful to do this when you are having the problem.

The power cycle that you did, while clearing the problem, also removed any evidence of what the problem might be.

Your main hopes short of a dump would be the OLDsyslog.log file, the nettl.log, or /var/tombstones/ts99.

Check the date INSIDE the ts99 file on your system and see if it is the same as when your system stopped hanging.

If there is not a valid timestamp in the ts99 then you'd need to wait until another failure occurred.

Best regards,

Kent M. Ostby

[srinivas_34]

Hi Kent,
thanks for your help and advice. i really appreciate for your valuable suggestions.

[srinivas_34]

Hi Kent,
Again my HP 9000/K360 hung up without communication.There is no console messages this time & also the system console terminal is hung up.
please advice me for any solution.

[Kent Ostby]

Put the switch key to "Service" ...

You SHOULD be able to type cntl-b> TC

If not, there is a button on the back of the k-box labled TOC or TC or "Transfer of Control" ... this is a recessed button ... use a pen to push this button and this will cause a reboot and save a crash dump.

From there, you will probably need to get HP support to troubleshoot the dump.

Best regards,

Kent M. Ostby

[srinivas_34]

Hi Kent,
can u please give me some more information regarding TC and coredump.

thanks inadvance for your reply

[Kent Ostby]

The TC will force a reboot of the machine and save a crash dump so that it can be looked at to determine the cause of the hang.

The crash dump is a copy of the image in memory and HP has tools that you can run against it to help determine the status at the time of the failure.

If it were a panic or self-reboot, I would ask you to post the information here as it can sometimes be a fairly straight forward task to match the panic to a known problem.

Hangs are different so it is usually quicker and easier to have HP read the dump for you.

Best regards,

Kent M. Ostby

[srinivas_34]

Hi Kent,
thanks for your valuable suggestions.
i tried to do the same procedure as u told.
i mean after the key in service mode and ctrl-B > TC
I GOT THE ERROR SAYING
"RUN 42F REMOTE: DISABLED IN ACTIVE MULTIPLE ACCESS FAULT " with some junk characters on the screen & the system is not rebooted.

i even tried hard TOC by pushing the toc button on the back of k-class but nothing happend.

[Kent Ostby]

Hmmmm .. I've seen this kind of thing a few times with a system that had a problem in HW between the I/O card and the disk drive that swap is written out to or simply in the core I/O board.

Are there any chassis codes ( like the C440 ) which are currently showing on the console.

[srinivas_34]

Hi Kent,
this time there is no console messages like C440.I even run "stm" and check the status of all the hardware devices connected to my mechine.
Since my mechine not allowed to do TC, I guess i dont have crashdump.
i am right now analysing oldsyslog,rc.log files , but on the rc.log file i have follwing entry
"Configure system crash dumps
Output from "/sbin/rc1.d/S080crashconf start":
EXIT CODE: 0
"
please advice me for any solution