Diagnostic login facility receiving excessive errors from I/O sub sys

[Josee Bourget-Thuma]

Hi,

I just patched my HP-UX 11.0 O/S and the messages I get at boot are as follows:

"-diagnostic login facility has started receiving excessive errors from I/O sub system.

- I/O error entries will be lost until the calls of the excessive I/O login is corrected

- if the diaglogd daemon is not active use the daeman startup comand in stm to start it

- if diaglogd is ative, use hte logtool utility in stm to determine which I/O sub system is login excessive errors."

Any suggestions please?
Thanks!
Josee...

[Steven E. Protter]

You have a disk or core io issue.

dmesg

check for lbolt or powerfail in /var/adm/syslog/syslog.log

SEP

[Kent Ostby]

Josee --

You have a problem with I/O probably disk or card . It could also be memory.

As was mentioned, you can look for an lbolt message in syslog.log or dmesg output.

If you see something in either that looks like: b->dev = 0x00d000 then you can probably find out which device is having problems.

Do:

ll /dev/dsk | grep 00d000
(for the example above) and that will help you isolate which disk is the problem.

Best regards,

Kent M. Ostby

[Josee Bourget-Thuma]

OK, I gathered that much and would love to run STM but how do I check all this if I don't even have a booted system? IE No command prompt?

Thanks Guys!

[Brian Watkins]

Since you mentioned that this is an 11.0 system, you should be able to access the GSP and see if it recorded any hardware errors in its error logs. As long as the server has power, you should be able to access the GSP, even if you don't have a login prompt for the system.

Just hook a serial console to the GSP console port, log in, and type "sl", then "e" (to view the error logs) and then "n" (no filtering).
That will give you a dump of the GSP error logs on the system.

You may want to have the Response Center on the phone when you are viewing these logs, as they can help you decode any hexidecimal fault codes recorded in the GSP logs. Keep an eye out for anything with an alert level of 6 or higher (6 - 9 usually indicate recoverable errors, 10+ usually indicate hardware failure is imminent or has already occurred).

Go to http://docs.hp.com and do a search for GSP Error Logs. On the search results page, click on the "Chassis Code to FRU Decode" link. That will bring up an example of viewing GSP error logs (it's for the rp54xx series, but the GSP uses the same basic format across the board.)

I guess your other option would be to boot into single user mode and try to run STM from there.

Good luck and let us know what happens!

Brian