SCSI Initiator

[MikeL_4]

Our company hardware group added a SCSI attached, what they call a Multipack to our rp3410 server that contains eight (8) disk. We have one that keeps failing and they are telling me:

"You might have to see about changing the initatior to something other than 8. The controller should be on 7, so you should be able to use 6."

How do you change something like this, never had to do it before...

[Steven E. Protter · Accepted solution]

Shalom,

Mike, you should be able to use defaults most of the time, but you may have run out of scsi ids.

Can you post ioscan or ioscan -fnC disk output to look at.

http://www.pimpworks.org/hp/hpuxtips.html
http://docs.hp.com/en/5187-2239/apbs05.html
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=7201

SEP

[Torsten.]

Better to post

# ioscan -fn

An initiator will/should never have an ID of 8, because this is the lowest priority on the bus.

[MikeL_4]

Attached is the ioscan -fn from the server...

The disk that keeps failing is at: c5t8d0

[Kevin Wright]

The scsi target device ID of your disk is 8, the initiator of each scsi chain will by default be 7. They are not one in the same....

typically the only reason to change the initiator is when you have a cluster using shared storage over scsi. The initiator on one of the nodes needs to be changed to 6, as you can't have 2 ID's of 7 on the same bus.

[Michael Lampi]

When you say it "keeps failing", exactly what do you mean?

Does the device work for a while, then stops doing data transfers without experiencing I/O errors??

Does the drive physically fail, as in stopping spinning, making loud noises, smokes, etc.?

Do I/O transfers to the drive fail with SCSI timeout errors?

Does the drive never work?

[MikeL_4]

Started with:

Disk at hardware path 0/4/1/1.8.0 : Software configuration error

Disk at hardware path 0/4/1/1.8.0 : Device connectivity or hardware failure

Disk at hardware path 0/4/1/1.8.0 : I/O request failed

Attempting to relocate subdisk ignitedg12-01 from plex ignitelv-02.
Dev_offset - length 54619320 dm_name ignitedg12 da_name c5t8d0.


When I tried to replace with a new drive, the new disk failed on initialization in Veritas...

[Torsten.]

I wonder about your device.

On other similar systems the SCSI controller appears as

ext_bus 4 0/2/1/0 c8xx CLAIMED INTERFACE SCSI C1010 Ultra160 Wide LVD A6829-60101

but on your system

ext_bus 5 0/4/1/1 c8xx CLAIMED INTERFACE SCSI C1010 Ultra Wide Single-Ended A6829-60101


Maybe at least 1 disk is downshifting the bus.


What gives

# scsictl -c get_target_parm /dev/rdsk/c4t8d0

I don't know about this disk devices, but an internet search results in many critical comments ...

[Torsten.]

And another point:

You say you have 8 disks.

There are 14 external disk device files.

Are the disks shared between the 2 HBAs?

In this case you must set one initiator to ID 6.

But where is the disk 8?

You have SCSI IDs 2 3 4 5 8 9 10 visible on each controller = 7 disks.

Please clarify!

[MikeL_4]

Sorry, there are 7 Disk on both c4... and c5...

The plan was to install the disk at c5... and mirror the data on the disk at c4...

scsictl -c get_target_parm /dev/rdsk/c4t8d0

TARGET LIMITS
-------------
flags: 0x0
width: 16 bits (8 = Narrow; 16 = Wide)
req/ack offset: 62
xfer rate: 80000000
SPEED: 160 MB/s (Ultra160 Wide)

NEGOTIATED TARGET VALUES
------------------------
flags: 0x0
width: 16 bits (8 = Narrow; 16 = Wide)
req/ack offset: 31
xfer rate: 20000000
SPEED: 40 MB/s (Ultra Wide)

[Torsten.]

To be honest - I would use these disk for my PC only and connect better and fully supported disk to the server ;-)
Look, they are slow down the bus. Google for these disks ...

Anyway, if they share the bus you must set one HBA ID to 6.

[MikeL_4]

Then back to the origional question:

How do you change the HBA ID to 6 ??

[Torsten.]

Here is the manual:

http://docs.hp.com/en/A6829-96002/index.html

[Torsten.]

If the disk keeps failing, you either have a bad disk or a bad slot.

In addition the disks are running at 40 MB/s only - this looks like a known issue for this kind of third party disks.

If you do some research you will see you are not alone ...

If you use this server for production, you should consider to use other disks ...

[MikeL_4]

Install of latest SCSi patch: PHKL_36521 resolved the issue..

Thanks

[MikeL_4]

.