Jamaica A3312A 2 D380 and testing clustering

[Ed Dubrovsky]

Hi All,

I am having some fun time setting up a test cluster. My setup is as follows:

HPUX11.0
QTY 2 - D380 HPUX 11 installed. A2487 fwd scsi cards in each server with yellow terminators installed on each card and each card has different ID.

Jamaica Unit with 4 9GB FWD disks in the unit. Cables from one port on the back to first D380 (lets call it node1) and cable from another port connected to second D380 (node2). 1 cable is connecting one port from each side of the jamaica to each other to create one scsi bus.

Observation 1 - running SEA on each machine shows the 4 9GB drives available on each machine (drives are visible with SCSI IDs 8,9 and 12,13 respec.).

Observation 2 - Booting up node1 completes with no problems. Node is up and running.

Problem - Booting node2 I get LCD init code CEF2 (some sort of IO error?) and system is pretty much stuck at loading the kernel.

Am I missing something obvious? What should I be looking for?

Any suggestions would be appreciated!

Thanks!
Ed

P.S. This thread has been moved from General to Disk Array. - Hp Forum moderator

[Joshua Scott]

Ed,

How are the internal disks configured on each? there is some dipswitches on the back of the hotswap drive backplane. make sure they are the same on each.

CEF2 means Kernel is about to configure the I/O. This is a normal part of the boot-up process. How long does it wait at this point?

-Josh

[Ed Dubrovsky]

Hi Joshua,

I have not looked at the dip switches on the D380. Both systems have 4.3GB drives if I recall correctly they are 4.5.0 and 4.8.0 on the internal bus for both systems. But that is a different bus then what the jamaica is connected on.

The systems boot of the internal 4.3GB.

CEF2 error goes for a long time...I did not check it this morning but yesterday it was not moving after about 1.5hours...

Thanks!
Ed

[Joshua Scott]

Ed,

Are both systems identical in every way? (all I/O card in the same slots on both D380s, same size drives in the same slots?)

I've seen this happen before on one of my D-class servers, and I beleive it was due to drive failure. you may have to get a replacement if that is the case.

Try booting it up standalone without the jamaica or any other externals attached.


A couple of threads on CEF2:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=224998

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=114538

-Josh

[Bernhard Mueller · Accepted solution]

Ed,

first, stop boot on node2, re-try sea, OK?

second, reverse the order: bring down node1, see if node2 boots, if so try to boot node1, see if same symptoms.

then you should be able to tell that it is either a node-specific issue or a setup problem.

If suitable try disconnecting the Jamaica from node2 and see if it boots.

I take it the FWD SCSI cards are *not* the ones that connect to the internal drive bay (the fixed targets of the slots would conflict with ID 8 and 9 in the Jamaica).

Next check SCSI cables and connectors for bent pins.

If you are still at a loss, something tricky: make sure the yellow terminator resistors on the SCSI cards *all* have the correct orientation (there is a mark on them)!

Regards,
Bernhard

[Steve Robinson_12]

Have I missed something here? I don't think you mension the scsi id of both controllers on the bus, default is 7 so if you have two scsi controllers on a shared bus you should set one to say default 7 and the other to say 6.

Confirm this by booting each node at a time and check the controller id with ioscan

[Ed Dubrovsky]

Hi,

I did mention that each card has a different scsi id but not the id #. Scsi cards have 7 and 6 as scsi ids. SEA also works from both servers. I have now figured out the problem. Each server has an additional 2 scsi cards installed which do not have terminators installed on them. For some reason the one D380/2 had no problem with that...while the other D380/1 would just sit at CEF2. Once I placed diff. terminators on the remaining cards (not being used) everything was cruising along perfectly.

Thanks for everyones help. Your suggestions gave me a few hints!

Now onto the cluster setup...

regards,
Ed