AIX

Hello guys,

on one of our AIX VIO LPARs I can not remove hdiskpower devices, which seemed to presented to the VIO in an incorrect way and behaviour (not all paths can be seen, pseudo device name is unknown etc.)

Therefor I decided to deattach the hdiskpower device from PowerPath controll by:

1. powermt remove <hdiskpowerXX> ---> did not work

then directly via:
3. rmdev -Rdl <hdiskpowerXX> --> i got this message:

rmdev: 0514-516 Device configuration database lock service timed out.
Please retry the command later.

3. Check of the config_lock led me to:

Code:

# fuser /etc/objrepos/config_lock
/etc/objrepos/config_lock:  15466568 16449626 20316396 21364886 23265418 23724286 29032682 30015688

4. Examing the PIDs by ps:

Code:

15466568      - 647:39 cfgpower
# ps -A|grep 16449626
 16449626      -  0:00 powermt
# ps -A|grep 20316396
 20316396      -  0:00 cfgmgr
# ps -A|grep 21364886
 21364886      -  0:02 chcod_chrp
# ps -A|grep 23265418
 23265418      -  0:02 ha_star
# ps -A|grep 23724286
 23724286      -  0:00 cfgmgr
# ps -A|grep 29032682
 29032682      -  0:00 cfgmgr
# ps -A|grep 30015688
 30015688      -  0:00 cfgmgr

Question:

Is there any way out of this mass, without rebooting the VIOS LPAR?

Sincerely

Tomek

Moderator's Comments:

edit by bakunin: please use CODE-tags for for commands, terminal output and file content you present. Actually this is a rule here and you have accepted it. Thank you.

kill -9 helps sometimes.

In principle, yes, but isn't that a little over-enthusiastic? I tend to try kill -15 always first and only if the process in question is poorly written (and doesn't honour the signal) use kill -9 as a last resort.

The difference is that -15 gives the process to be ended a chance to clean up after itself: remove temporary files, remove allocated shared memory segments, remove opened sockets, ..., whereas -9 is a - to the process external - headshot.

Its suicide versus kill and, as we all know since the days of M.A.S.H, suicide is painless ... ;-))

I hope this helps.

bakunin

Additional information on that:

in the HMC I see the reference code 0999 (Disk Array Subsystem being configured)...

Yes - and several of your configuration processes already hang. Your perhaps best option is to reboot the system completely and take it from there. If even the cfgmgr already hangs with several instances there is something seriously wrong and i don't know if you are ever going to recover from that. A reboot will at least bring the system into a defined state.

I hope this helps.

bakunin

Hey,

thx to all who gave constructive explanations and hints...

I decided to reboot all affected VIOS -> did work.... :-)