Power Button - Chass cntrl

[Hiroshi Leong]

Sometimes it takes some time until the server actually shuts down.

Below are the BMC log taken from the node after shutting down the server. The first action for pressing the button is recorded in #0010, however, the CPSB didn't go off at this point. I didn't release my finger and kept pressing the button, then the CPSB finally was shut down . I believe this actually occurred in #0030.

There was one event observed between two 'Power Button Pressed' events, which was 'Power Button Chass cntrl', and I was wonding what this was about. Do you know how to interpret this?

cli>sl -e
# Sev Generator/Sensor Description Event ID Data, Timestamp
---- - ---------------- ------------ ----------- --------------------------
0010 - Power Button Pressed 04-14:6F:00 2006-09-21 18:18:27
0020 - Power Button Chass cntrl 04-12:70:A3 0E:04 2006-09-21 18:18:30
0030 - Power Button Pressed 04-14:6F:00 2006-09-21 18:19:00
0040 - BMC Chass cntrl 00-12:70:A3 81:06 2006-09-21 18:19:36
0050 - BMC LPC reset 00-12:70:02 2006-09-21 18:19:40
0060 - SFW Boot start 00-1D:0A:00 2006-09-21 18:19:43
0070 2 CPU0 Boot start 00063 DT 14 0000000000000000
0080 2 CPU0 Boot start 00063 Time 2006-09-21 18:19:43
0090 - BMC LPC reset 00-12:70:02 2006-09-21 18:19:17
00A0 - SEL Time Set Set FD-C0:03:01 2006-09-21 18:19:17
00B0 0 CPU0 002C5 DT 16 0000000000000000
00C0 0 CPU0 002C5 Time 2006-09-21 18:19:20
00D0 - SEL Time Set Set FD-C0:03:01 2006-09-21 18:21:07
cli>IPMI 28 40
00 01 0D 00 20 3F 13 D8 12 45 4F D5 12 45 03
cli>exit;

If the CPSB immediately goes off after pressing the power button, the log looks like below...

cli>sl -e
# Sev Generator/Sensor Description Event ID Data, Timestamp
---- - ---------------- ------------ ----------- --------------------------
0010 - Power Button Pressed 04-14:6F:00 2006-09-21 16:58:36
0020 - BMC Chass cntrl 00-12:70:A3 81:06 2006-09-21 16:59:42

Can you help?

Regards
Hiroshi

[Luk Vandenbussche]

Hiroshi,

What model of server do you have?

[Hiroshi Leong]

Hi,

I have a cx2620 server.

[Sameer_Nirmal]

Looking at the timestamps of both the logs, I see there is a difference of 3 secs ( buttom press to bmc's power off). The difference is justified on account of special power button funcationality built in and pressing button for more than 3 secs.

Refer the link at
http://docs.hp.com/en/AB587-96006/ch06s01.html#babhgfdd

Now looking at the events in the first log, the event " Power Button Chass cntrl " I would say it signifies the sensed chassis control event which reached/passed the internal initial time slot of 1-3 secs. The power button sensor again sensed the button press as next event ( maybe the switch tweak if not your finger to blame ) before BMC make chassis power off.

You can get more details of the events by selecting the "text" mode in MP -> SL -> E

[Hiroshi Leong]

Hi guys,

Thanks for looking into this for me, but something still doesnâ t make any sense. The way the server switches off is inconsistent.
Holding the power switch for 10secs will cause the CPSB to power off during this interval. However at other times the server will only turn off after you hold the button for 10sec and then release the switch.

The power switch is pressed for 10sec then released. The CPSB powers off, generating this log file:-
Power Button Pressed 2006-09-27 11:32:44
Power Button Chass cntrl 2006-09-27 11:32:52
BMC chass cntrl 2006-09-27 11:33:52

The power switch is pressed for 10sec. During this time the CPSB power off the I release the switch:-
Power Button Pressed 2006-09-27 11:39:10
BMC Chass cntrl 2006-09-27 11:39:56
BMC LPC reset 2006-09-27 11:39:59

Any ideas on this?

Regards
Hiroshi

[Sameer_Nirmal]

Can you post the Forward Progress Log (FPL) in both inconsistent scenarios?

cli> sl -f
or
MP -> SL -> FPL ( text mode )

What's the firmware version of the server?
cli> sr
What you get with
cli>pc
cli>pr
cli>ps

Do you hear a beep while button is pressed in either scenario?

The significant difference between the two scenarios lies being having ChassisControl event logged by sensor PowerButton.

1F20 - Power Button Chass cntrl 04-12:70:A3 00:04 2006-09-27 11:32:52

With the latest BMC firmware, this is supposed to happen. The inconsitency might be because this event is not getting logged in the second scenario causing unit getting power off only after releasing the button.

[Hiroshi Leong]

Hello,

Thanks for looking into this. The first log you requested is not available on the menu.

cli>pr
Power restore policy is Always on
cli>IPMI 28 40
00 01 26 02 90 1D F3 9E 1B 45 7C 6E FE 44 03

cli>ps
System power is on
cli>sr

Current firmware revisions:
MP FW E.03.15e
BMC_FW 03.56
EFI FW 05.14
System FW 01.10

USING PC CONTROL OFF
cli>pc off
System will be powered off
Confirm? (y/[n]):y
cli>pc on
System will be powered on
Confirm? (y/[n]): y
cli>pc off
System will be powered off
Confirm? (y/[n]):y
LOG:
2270 - BMC Chass cntrl 00-12:70:A3 80:06 2006-09-28 10:16:59
2280 - BMC Chass cntrl 00-12:70:A3 81:06 2006-09-28 10:17:06
2290 - BMC LPC reset 00-12:70:02 2006-09-28 10:17:10
22A0 - BMC Chass cntrl 00-12:70:A3 80:06 2006-09-28 10:17:11


Regards
Hiroshi

[Sameer_Nirmal]

Well, the FPL in both the scenarios would help to track the events and to know the cause of inconsistent behaviour of power button funcationality. You could have collected the FPL from MP as I mentioned before.

Anyways, the tracking the event through FPL won't fix your problem and we know the Power Button Chassis control event is missing sometimes. This is most probably the cause of inconsistent behaviour.

As we can see, this funcationality ( power button ) is acheived in the firmware. Hence the issue with the existing firmware can't be ruled out. I would suggest to upgrade the server fimrwares to latest available at

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15351&prodSeriesId=1816642&prodNameId=1816650&swEnvOID=54&swLang=13&mode=2&taskId=135&swItem=ux-43381-1

The latest firmware should resolve the problem.