K-460 and A3312A Storage Enclosure

[sunil kattikar]

Hi,
I am having one K-460 along with one A3312A Storage Enclosure. I have been charged with upgradition of server by adding new A3312A with 8 no of 4.3Df disk. I have added additional F/W SCSI ctrl(A2969-60001) on HSC Expantion board and added this new box on this controller. At PDC prompt I am able to detect all disks. But after booting it goes upto hpux(10/0.6.0;0)/stand/vmunix and then it gives next line and after words it gives me system panic. A system crash has occured. When system halts for some time at above kernel path on my new A3312A box HDD LED glows continuously for one disk(A1) and for rest of the disk it flashes 2/3 timesand goes off . Then it Continuously glows for Disk in A2 slot and for rest of the disks it flashes and goes off. Same for A3 Disk and then it gives system panic. My H/w path are

Path Before After
p0 10/0.6 ( boot disk ) 8/0.10
p1 10/0.5 8/0.9
p2 10/0.4 8/0.8
p3 10/0.3 8/0.5
p4 10/8.10 8/0.4
p5 10/8.8 8/0.3
p6 10/8.6 8/0.2
p7 10/8.5 8/0.1
p8 10/8.4 10/0.6
p9 10/12/5.2 10/0.5
p10 10/12/5.0 10/0.4
p11 10/12/6.0 10/0.3
p12 10/8.10
p13 10/8.8
p14 10/8.6
p15 10/8.5
p16 10/8.4
p17 10/12/5.2
p18 10/12/5.0
p19 10/12/6.0
As my path P0 is getting changed so at PDC I tried with boot P8 ( new P0) and also boot 10/0.6 but still it was giving me system panic. After tring a lot I have connected this new A3312A in cascade with the existing box i.e. all 13 disks ( 5 exsisting and 8 new ) came on single controller 10/8.1-14 ( 7 assigned to controller ) and by doing this My path P0 remain same and system started booting normally.

Now my question is where this P0 is playing a role. If respestive hardware path is getting changed then why OS gives system Panic. Where OS refers then path?

Pls. note all my disk are OK. SCSI termination is proper and also all SCSI cables are also ok.

[Sudhakar.J]

Hi ,
As per the above list, H/W path is getting changed when u connected your additional A3312A box is connected to another controller, which may not work since /etc/lvmtab refers to old device path(with previous H/W path).

Could u try by connecting old A3312A box to the new controller and the new A3312A box to old controller.So that u may get the order in which OS disk comes back to 10/8.4 10/0.6 .

ATB,
Sudhakar

[Brian M Rawlings]

One question: do you have any internal drives in the K460? The K-class servers have a built-in FWD controller which has a port on the Multifunction card, so that you can connect external FWD devices (like the HASS/Jamaica chassis you are connecting).

If there are any internal drives, however, they take up SCSI IDs on that internal FWD bus, typically 6, 5, 4, 3 (traditional boot drive in HP is SCSI ID 6, others may be there or not, but go in that order if they do).

So, you may have a SCSI ID conflict. On the HASS/Jamaica (A3312A), the SCSI IDs are DIP switches on the back of each drive bay, with the ID set in binary.

To see if you have internal K460 drives, you have to open the front door. Below the CD & DAT bays is a black metal grill, inside it you can make out the drives, if they are there. There can be up to four drives, generally starting at the top.

You say that your drives, cables, and terminators are all OK, but something is causing this. I would try the following:

1> leave everything cabled up as you have it, but pull all of the new drives out an inch or so (disconnecting them). This will keep the bus terminated, just as it is now, but will eliminate all the new drives from the equation. See if you can boot w/o panicing.
2> If no panic, add the new drives back in, either one at a time, or half at once. You can eliminate half at a time if you begin to panic again, and quickly work out which drive it is, if that turns out to be the cause.
3> Disconnect the new chassis, and rotate pairs of the new drives through the empty bays in the existing chassis, which has presumably worked OK for you. You'll need to double-check for unique SCSI IDs, since those 3 bays haven't been used in the past. The bays have the SCSI IDs, not the drives...
4> The new FWD controller, A2969A, has terminating resistors on it, or it should. Careful, these might have been removed (for multi-initiator SCSI configurations like MC/Serviceguard with 3-4 nodes), or they just might have been left off. You need them, they are generally a pair of yellowish DIP packages near the HD68 connector...
5> You have duplicates of all the gear; cables, terminators, drives. If there was ever a situation where you could troubleshoot by swapping out suspect items, you're in it.

One thing, the very low level 'SEArch' function at the boot prom menu is not a perfect test of the disk infrastructure. It is a good sniff at it, and will often fail when it should, providing valuable insight. But it can succeed where the OS will not, due to timing, or much more robust I/O, or whatever. So, just because the boot prom/PDC level can find drives OK doesn't mean that all is well in the SCSI chain.

Later... --bmr

[Steven E. Protter]

You might have the cabling to the controller card wrong. I was fiddling with a D box getting ready for decommission. It has two scsi cards and a cable connected to one. There was a connector hanging off the second hard that looks like it should have been plugged tot he scsi card.

Four hours later, I realized that no, I should not have connected that cable.

As probably noted earlier, check the SCSI ID, cables and connectors.

Try disconnecting the enclosure, booting the box.

Then connect the array to the scsi card and run ioscan.

It should still show up.

There may be drivers you need to add to the kernel.

It is possible that in transit a disk failed, but that should not result in a kernel panic.

SEP