Warning light on pseries.

[blasvinas]

Hi,

My company have several pseries servers (p670, p650, and p615). These servers always have the warning light on. I execute diag and chose set the Attentions Indicator to Normal and turned the light off, but when we reboot the server the light is on again. This is happening with all the pseries servers. I download the lastes microcode for one of the p615 but the problem still continue.

Somebody have the same problem? Do you know how fix it? Thanks in advance for your cooperation.

 

[mrberry]

I had this happen to me upon boot.

Check you errpt for any hardware errors.

You might want to do an errclear 0 before a reboot so that you cut down on the errpt messages you have to look for - but remember this will erase your errpt history.

Mine was due to a tape device.

Hope this helps.

 

[DukeSSD]

http://publib16.boulder.ibm.com/pseries/en_US/infocenter/base/hardware_docs/pdf/380509.pdf

The system attention light comes on because the system has a problem.
Turning off the light is not the answer.
The first step it to look at the errors posted in the error report just after the boot.
If you don't understand the errors (some are not at all easy to understand). The next step is to run diagnostics.
The diag command will start the diagnostic program but you will need root authority to run diag.
Hit enter when you have "read" the licence page then select advanced diags then select the problem determination option. This should be ok to use during normal run time because it will analyse the error report before giving you the option to actually test any hardware.
It will probably respond with something like "A problem was detected....... (round about now, because it is now that you are running the program) and should give the AIX location and FRU (part number) of the device with the problem, it will also give the sequence number which relates to a specific error report entry.
It will then ask you to hit enter to continue and show the next problem.
When all the error report entries have been analysed it will present a resourse list, this gives you the option to actually test some hardware. This is also usually OK to do during run time but if the device is busy, ie: network card or hard disk, etc. it will need you to free the resourse before it can fully test it.(obviously not always possible on a live system)
You can use the book linked above, follow the start of call map, and also dig out the specific service guide when required:

http://publib16.boulder.ibm.com/pseries/en_US/infocenter/base/hardware.htm

These should point you to the real problem.
If you have IBM hardware support just place a call.
If you have third party hardware support you may want check if you are charged by the call, if so it is probably worth "walking the maps" to get a good idea of what the problem really is before getting a bill for it.

 

[Yegolev]

there is a firmware fix to an always-on check light. may as well update firmware since IBM will tell you to do that and call back anyway.

IBM Certified -- AIX 4.3 Obfuscation

 

[mjldba]

in addition to reviewing the errpt & running diagnostics, take a look at the bootlog in /var/adm/ras if the attention LED comes on immediately after boot.