External HA disk enclosure A3312A connected to a V2500

[pj_11]

Hello,

Can any one help me i currently have a V2500 With one external disk enclosure attached.This is used to boot the V2500.The customer have added a second disk enclosure however one bus
can only be seen.Both buses connect to different scsi cards on different pci card cage.The card is a A4800A.is ther a limitation
on how many disk you can connect.Bearing in mind they are attached to two different scsi cards in seperate PCI card cages.

Thanks,

PJ

[Florian Heigl (new acc)]

The first thing that came to my mind would be a scsi address conflict, i.e. one of the controllers 'sharing' it's address with one of the disks - if termination and everything look ok, that's where I'd start searching.

[Stefan Stechemesser]

Hi,

you can connect up to 15 disk to a A4800 card, but each one needs to have a unique SCSI ID.
Please check if the SCSI busses are terminated properly. The A4800A card needs a terminator to be plugged on the internal port (Part Number 1252-6520). And of cause ther also have to be a teminator on the enclosure, too.
Maybe you could add an "ioscan -f" output of the system, then we can see if the A4800A card itself is recognized by the kernel.

best regards

Stefan

[Andrew Rutter]

hi,

There was two revisions of the A4800A i/f's originally.

As an ordering option you just ordered A4800A for the V class systems but you had to order /002 for the RP's.

The option /002 gave you a terminator which was needed to terminate the int bus on the RP's. I dont think you need to do this in the V classes.

Also the V class A4800A's had the A4800-62001 part number on them
The A4800A/002 had A4800-62002 part number on them.
There could be slightly different firmware on the cards?

Check the part numbers on the cards.

Then you could check the scsi id of the card at BCH. boot the system without the hass box attached and check if its seen at BCH.
This rules out scsi id conflicts with disks and cable issues.
The V class supports upto 22 cards aswell.

There is also numerous docs at

www.docs.hp.com

for the installation of the card which may help you.
Its more likely to be a scsi id conflict but the others are worth checking.

Andy

[Andrew Rutter]

hi you could check this patch aswell

http://www4.itrc.hp.com/service/cki/patchDocDisplay.do?patchId=PF_CV260720

Andy

[pj_11]

Hello All,

Thank you very much for your input.There is a powerdown of the datacentre this weekend.So
i wil try the things you have said.I have connected the disk enclousre to a D class it
all works fine.So it looks to me like a trouble shooting job on the V class.Could
be cable,card,termination,card cage.Sorry for
the late response i got caught up in a nother
problem.

PJ

[timmy b.]

Hello, PJ.

To this point everyone has been thinking from the server end, so I'll go the other way and think about the Jamaica enclosure. (That's what HP called the A3312A)

On the back of the enclosure there are 8 little windows with DIP switches. This is were you set the SCSI address for each individual drive slot. They are simple binary switches. You don't want any two of them set the same on a bus (A or B) if you are connected to separate controllers, or if the A & B busses are cabled together you don't want any duplicates in the entire enclosure. If you don't understand the switch settings, reply and I'll post a breakdown on the SCSI ID vs. switch settings.

Most often the controllers are set to SCSI ID 7, which is the highest interrupt, so you don't want one of the drives set to that, either. In a dual-controller environment the second controller is typically set to 6, this would be a ServiceGuard configuration, and you wouldn't want to use that setting either.


I hope this all helps!

Good Luck - Tim

[Dave Unverhau_1]

PJ,

I might as well throw in my .02 worth, as well...

When things are shut down, you might want to check the pins on the SCSI cables. As I recall, the SCSI connectors on the A3312's were recessed a bit into the rear bulkhead, and it was pretty easy to misalign the connector slightly and bend or break a pin while trying to connect things.

Also, I remember a service note coming out several years ago that dealt with the SCSI cabling on Jamaica enclosures. The problem was with using cables that were too SHORT. Originally, the two sides of the jamaica enclosures were tied together with a 1/2 meter cable (maybe it was shorter), and often the enclosures were connected to the hosts or daisy-chained with 2M or shorter cables. The problem was a data-race issue, and the fix was to ensure that the two sides of the enclosures were connected by a cable that was at least 1M in length and that they were daisy-chained or connected to the HBA's by cables at least 2M in length.

I'm not saying you have to do this, but if nothing else works and you have very shor cables, you might want to keep this in mind.

Good Luck!

Dave

[DCE]

PJ

One more item on the A3312. It can accomodate up to eight disks and up to two channels. BUT a common configuration was to put the disks all on one channel. This was done with a very short external SCSI cable that went from one side of the box to the other.

Dave

[pj_11]

Hello All,

Just to let you the terminator was not installed on the internal connector of the scsi card.This caused the disk on the bus not to be seen.


Thank you

PJ