Old error messages appear on a Console at shutdown time.

[中島 正春]

System: SuperDome/HP-UX11.11

I sometimes get old error messages on a console at shutdown time as follows.

SHUTDOWN PROGRAM
06/12/03 19:47:49 JST
...
reboot: CAUTION: some process(es) wouldn't die
6/0/14/0/0: Unable to access previously accessed device at nport ID 0xdc.

The above error is old error messages appeared on the console(also in syslog) as follows.

May 31 03:30:13 WINAP01 vmunix: 6/0/14/0/0: Unable to access previously accessed device at nport ID 0xdc.

It seems me that the above old messages remaind in a some kind of kernel message buffer.
And HP-UX displays the buffer(old error messages) at the end of shutdown time.

Is it true?
Are there any patches for this problem?

Best Regards,

Masaharu.

[Tom Geudens]

Hi Masaharu,
These messages indicate that you've changed the AL-PA address of a device on your loop (or changed the device and the new device has a different AL-PA). Did you ? The messages are very persistent and will NOT go away without you taking action. Here's the commands that give you the needed info and can fix the situation (note that these are commands in my environment, the "values" in yours will differ) :
#tdlist
-> will give you the AL-PA addresses
#fcmsutil /dev/td1 replace_dsk 0xdc
-> will replace the AL-PA address for a given device

Hope this helps,
Tom Geudens

[中島 正春]

Hello Tom,

Thank you for your answer.
And I am sorry for my less explanation.

Actually, we fixed the problem of the error messages before booting.
And after that I did not get any error messages for this problem till the system shutdown.

I have another example as follows.

--messages appeared when the system lived--

Jun 12 21:32:27 WINDB01 vmunix: The diagnostic logging facility is no longer receiving excessive
Jun 12 21:32:27 WINDB01 vmunix: errors from the I/O subsystem. 7 I/O error entries were lost.
...
Jun 18 22:30:06 WINDB01 vmunix: bindresvport:loop_twice port assigned 512



--messages at shutdown time--
*** Time-Stamp;030720/100002 ***
Shutdown at 10:28 (in 0 minutes)
...
Console Login: DIAGNOSTIC SYSTEM WARNING:
The diagnostic logging facility is no longer receiving excessive
errors from the I/O subsystem. 7 I/O error entries were lost.
bindresvport:loop_twice port assigned 512
sync'ing disks (4 buffers to
Closing open logical volumes...
Done
Boot device reset done.


I think the above error did not occur at shutdown time(030720).
I think kernel just displayed from its messages buffer when it dies.

Because
1.I think the error in the last shutdown
As you can see, the messages appeared at the almost end of shutdown messages.
I think at the time, DIAG daemon would die.

2.In the above examples, two error messages displayed at the shutdown time.
Those messages are exacly same messages as the messages occured when the system lived.

3.When I get some critical messages(DIAG, driver error, etc) during the system working, I always got the exactly same messages at the end of shutdown time.

Regards,

Masaharu.

[Steven E. Protter]

I'd check for address mismatches et al. Did you change the scsi id of a card?

reboot: CAUTION: some process(es) wouldn't die
That message is not very serious. Check your shutdown scripts and make sure they are following application guidelines

Check the shutdown log for problems.

SEP