iLo2 poweroff problem

[Jure Pecar_1]

Hi all,

I have a pair of DL385G2 and I want one to be able to shutdown the other. They will be used in linux/heartbeat setup, so I need stonith to work.

Poweron/off/reset works beautifully if I use iLo web ui. However, when I use heartbeat riloe stonith plugin, it does not work. Debugging the problem, I ended with HP example Set_Host_Power.xml, which also does not work. iLo accepts the request to poweroff the server, which is noted in the iLo logs, but the server remains running as it was before.

Here's the complete output of that script:

[root@hostA ~]# perl locfg.pl -s hostB-rm -f Set_Host_Power.xml


STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>
Scripting utility should be updated to the latest version.




STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>




STATUS="0x0000"
MESSAGE='No error'
/>


hostB-rm is the iLo IP of the other machine.

It is obvious that iLo reports no errors, but still the server is still running. This is the same even after I upgraded iLo2 from 1.29 to 1.30. I'm affraid I can't debug this further, as the problem seems to be between iLo and the machine. But using the Power Management in web ui works as expected ...

Any ideas?

[Jure Pecar_1]

As soon as I got a hint on linux-ha mailing list that SET_HOST_POWER sends an ACPI event, I did a mental connection and found out that acpid was not running. After I enabled it, this started working as expected.

[acartes]

The set host power messages send virtual power button presses, which are interpreted by the host as ACPI events which in turn should initiate a graceful OS shutdown.

There are other RIBCL scripts that you can use. Consult the scripting guide:

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00294268/c00294268.pdf

And look for info on:
SET_HOST_POWER
RESET_SERVER
PRESS_PWR_BUTTON
HOLD_PWR_BUTTON
COLD_BOOT_SERVER
WARM_BOOT_SERVER