FC60 LUN Disappearance on Controller

[Sunit Shukla]

Hi Guys !

FC60 Disk Array is connected with RP8400 Server. Fan and Power Unit were changed and then RP8400 Server was rebooting, this created unique problem -> LUN Capacity did not show right on one controller. Once the controller was reset, LUN capacity came back. Some people say, it was because of firmware in FC60 (current one is HP10). It should be HP12. If we do not upgrade, this problem can happen again on any hardware replacement on server. I want to confirm that firware is the only reason for this kind of problem or any other reason. Your suggestion will be appreciated.

[Eugeny Brychkov]

None will confirm not seeing 'amdsp -a', 'amlog' and FC60 serial debug logs first.
You should upgrade to at least HP11.
BTW, where fan and power unit were replaced - in FC60 or RP8400?
What was exact problem - 'LUN Capacity did not show right on one controller'? Did you use 'diskinfo -v' and compared between different FC60 controller device paths? Was host unable to access data of the LUNs?
Does host have AM60Srvr software installed and running? Did you watch FC60 logs (amlog)?
You know even very very fault tolerant system will crash if you'll not notice one-by-one component failures.
I do not believe this issue you had is because of fan and PSU replacement. It happened before, but host noticed it only after reboot/rescan.
So the conclusion for you is to upgrade to HP12, install AM60 Server and montitor 'amdsp -a' (general FC60 state) and 'amlog' (FC60 logs)
Eugeny

[AKIRA ONOMOTO]

Thank you for responding, Eugeny. I'm a co-worker of Sunit. I would appriciate if you could give me more information on this issue. Actually we replaced fan and power unit of the server rp7400(not rp8400), then we rebooted the server. After that we first had problem. We, of course, have AM60Srvr on the server and did run 'amdsp' to check disk status. The result was like the attached. 6 of 10 LUNs were recognized with the status of "not responding", though disk were working and accessible normally. In fact, Oracle database, which is constructed on FC60, could be used without problems. We can say that the result of "amdsp" was different from the real status of disks. As of now, we had only the informaiton that it(wrong display) was possibly due to the old controller firmware version. If you are familiar with this problem or have had the same problem before, please give us informaiton.
In addition, We are worried about having other more serious problems if we continue to use the old firmware HP10 and HP11 that are used on our two FC60's respectively. If you know something about the risk of the old firmware, please help us with informaion.

[Eugeny Brychkov]

Akira,
please see my previous post. Amdsp output is really incorrect, although hardware components states are GOOD/OPTIMAL.
But I would not give any recommendations and develop actions plan here because I'm lack of information (serial debug logs).
I strongly recommend to do the following: a) make FC60 full data backup and b) call HP to fix FC60 and upgrade firmware.
Maybe simple FC60 power cycling (with correct order: OFF: stop host activities [unmount/deactivate], wait until FC60 yellow cache led will go off, power off MAIN controller enclosure and only then disks; ON: power up disks, wait for 30 seconds, then power up main controller enclosure) will solve the issue, but I would not take responsibility for it. That's why I'm asking you to call HP - to have everything analyzed and action plan with contingency developed.
Good luck!
Eugeny