KCLASS giving SCSI: Read error

[Binu_2]

hi all,
We are sharing RA 8000 storage with HP900 K580,2*L1000, 1*IBM Netifnity (windows 2000)
servers connected through COMPAQ SAN HUB.
Whenever the IBM server is rebooted or scanned for disks the K580 gives

Jul 14 01:42:35 vmunix: SCSI: Read error -- dev: b 31 0x067600, errno: 126, resid: 2048,
Jul 14 01:42:35 vmunix: blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
Jul 14 01:42:35 vmunix: LVM: vg[1]: pvnum=0 (dev_t=0x1f067600) is POWERFAILED
Jul 14 01:43:01 vmunix: LVM: Path (device 0x1f067700) to PV 0 in VG 2 Failed!
Jul 14 01:43:01 vmunix: LVM: vg[2]: pvnum=0 (dev_t=0x1f067700) is POWERFAILED

on both LUNS assigned to the K580.But other
L1000 servers works without any problem


Has anybody faced such problems?
Is there any compatibiltiy problem with the Storage and Intel H/W?



************************RA 8000 details***

HSG80_UP>show this
Controller:
HSG80 ZG85204481 Software V86G-9, Hardware E03
NODE_ID = 5000-1FE1-0000-71C0
ALLOCATION_CLASS = 0
SCSI_VERSION = SCSI-2
Configured for MULTIBUS_FAILOVER with ZG85204657
In dual-redundant configuration
Device Port SCSI address 7
Time: 29-JUN-2004 19:48:33
Command Console LUN is lun 0 (NOIDENTIFIER)
Host PORT_1:
Reported PORT_ID = 5000-1FE1-0000-71C3
PORT_1_TOPOLOGY = LOOP_HARD (connection down)
PORT_1_AL_PA = 71 (not yet negotiated)
Host PORT_2:
Reported PORT_ID = 5000-1FE1-0000-71C4
PORT_2_TOPOLOGY = LOOP_HARD (connection down)
PORT_2_AL_PA = 72 (not yet negotiated)
NOREMOTE_COPY
Cache:
256 megabyte write cache, version 0012
Cache is GOOD
No unflushed data in cache


regds
Binu.K.V

[Kyle Martin]

Binu,
Here are some ideas to see if you have a bad disk or a bad connection, which is most likely what is happening.

Run this command:

# dd if=/dev/dsk/c#t#d# of=/dev/null bs=1024k

Let it run. You can ctrl+c out of it at any time. If it is a good disk or connection, then it should come back saying: ####+0 records in ####=0 records out. The #### should equal the same number. ANYTHING else, including it hanging or showing nothing, means that there is either a bad disk or a bad connection. Since the L class shows the disk as fine, then it would most likely be a bad connection. Check the HBA card with ioscan -fn, as well as the cables (physically). This is most likely the reason why you are getting these messages.

Hope this helps!

[Shaikh Imran]

Hi,
I have never used a san switch for hetrogeous environments But i can suggest you some check points.
If proper zoning is done this shouldn't happen.
No idea about what switch you are using.
Try using amdsp (see man amdsp ) command on k-class server and see what are the disk status.
Run fcmsutil command to see the link status.
Also you can always use STM for more detailed information and try verify in STM if any doubt.

Also if possible try to eliminate the K class and see the effect of scan on
L-class servers.

Also the message which you are getting is
pointing to a powerfaied disk.
pls identify the disk and run the above suggested test with dd

Regards,

[Binu_2]

hi all,

# dd if=/dev/rdsk/c6t7d6 of=/dev/null bs=1024k >/test/c6t7d6_readlog
60751+1 records in
60751+1 records out
# dd if=/dev/rdsk/c6t7d7 of=/dev/null bs=1024k >/test/c6t7d7_readlog
34714+1 records in
34714+1 records out
# cd /test
# ll
total 0
-rw-rw-rw- 1 root sys 0 Jul 13 17:58 c6t7d6_readlog



-rw-rw-rw- 1 root sys 0 Jul 13 18:39 c6t7d7_readlog
drwxr-xr-x 2 root root 96 Jul 13 17:56 lost+found

I got Empty files as output to these dd commands.

Since it is connected through SAN HUB, no zoning is done on this HUB. I believe Zoning is supported only in SAN switches?Right?