Unsolved
This post is more than 5 years old
5 Posts
0
5893
February 25th, 2005 11:00
SCSI errors with PV220s and 2x PE1750 (RedHat Linux)
Hi,
we have the following system set up:
2 x Dell PowerEdge 1750 (2x XEON 3GHz) with integrated(=onboard) LSI MPT U320 dual channel SCSI controller running RedHat Enterprise Linux 3 AS U2.
The first controller of each machine is used for the three internal hard drives which run fine.
The second controller of each machine is externally connected to a Dell PowerVault 220s containing 4 x ST373453LC drives (Seagate 73GB 15k U320).
Single node access seems to work without problems (partitioning etc.), however, after creating OCFS (ocfs-support and -tools V1.0.10-1, ocfs driver V1.0.12-1) partitions (for Oracle RAC), the following errors are written to the system log:
---< snip >---
Feb 22 10:15:16 durpdbc1 kernel: I/O error: dev 08:31, sector 47
Feb 22 10:15:16 durpdbc1 kernel: scsi1 (0,0,0) : RESERVATION CONFLICT
Feb 22 10:15:16 durpdbc1 kernel: SCSI disk error : host 1 channel 0 id 0 lun 0 return code = 18
---< snip >---
---< snip >---
Feb 22 10:15:16 durpdbc1 kernel: SCSI Error: (1:1:0) Status=02h (CHECK CONDITION)
Feb 22 10:15:16 durpdbc1 kernel: Key=6h (UNIT ATTENTION); FRU=02h
Feb 22 10:15:16 durpdbc1 kernel: ASC/ASCQ=2Ah/01h ""
Feb 22 10:15:16 durpdbc1 kernel: CDB: 28 00 00 00 00 69 00 00 24 00
Feb 22 10:15:16 durpdbc1 kernel:
---< snip >---
---< snip >---
Feb 22 10:21:41 durpdbc2 kernel: SCSI Error: (1:0:0) Status=02h (CHECK CONDITION)
Feb 22 10:21:41 durpdbc2 kernel: Key=4h (HARDWARE ERROR); FRU=03h
Feb 22 10:21:41 durpdbc2 kernel: ASC/ASCQ=40h/01h ""
Feb 22 10:21:41 durpdbc2 kernel: CDB: 28 00 00 00 00 69 00 00 24 00
Feb 22 10:21:41 durpdbc2 kernel:
Feb 22 10:21:41 durpdbc2 kernel: SCSI disk error : host 1 channel 0 id 0 lun 0 return code = 28000002
---< snip >---
These errors appear on both nodes for all disks regularly.
We have a similar cluster up and running which uses exactly the same hardware for the server machines but a different external scsi storage (simple "box" containing 6 x Fujitsu MAS3367NP without backplane).
Both clusters are configured the same way (secondary cluster option enabled in SCSI BIOS, different SCSI IDs for secondary controller). May assumption is that the problem is based in the external SCSI storage backplane as this seems to be the only difference.
Maybe anyone has made the same or a similar experience?
2 x Dell PowerEdge 1750 (2x XEON 3GHz) with integrated(=onboard) LSI MPT U320 dual channel SCSI controller running RedHat Enterprise Linux 3 AS U2.
The first controller of each machine is used for the three internal hard drives which run fine.
The second controller of each machine is externally connected to a Dell PowerVault 220s containing 4 x ST373453LC drives (Seagate 73GB 15k U320).
Single node access seems to work without problems (partitioning etc.), however, after creating OCFS (ocfs-support and -tools V1.0.10-1, ocfs driver V1.0.12-1) partitions (for Oracle RAC), the following errors are written to the system log:
---< snip >---
Feb 22 10:15:16 durpdbc1 kernel: I/O error: dev 08:31, sector 47
Feb 22 10:15:16 durpdbc1 kernel: scsi1 (0,0,0) : RESERVATION CONFLICT
Feb 22 10:15:16 durpdbc1 kernel: SCSI disk error : host 1 channel 0 id 0 lun 0 return code = 18
---< snip >---
---< snip >---
Feb 22 10:15:16 durpdbc1 kernel: SCSI Error: (1:1:0) Status=02h (CHECK CONDITION)
Feb 22 10:15:16 durpdbc1 kernel: Key=6h (UNIT ATTENTION); FRU=02h
Feb 22 10:15:16 durpdbc1 kernel: ASC/ASCQ=2Ah/01h ""
Feb 22 10:15:16 durpdbc1 kernel: CDB: 28 00 00 00 00 69 00 00 24 00
Feb 22 10:15:16 durpdbc1 kernel:
---< snip >---
---< snip >---
Feb 22 10:21:41 durpdbc2 kernel: SCSI Error: (1:0:0) Status=02h (CHECK CONDITION)
Feb 22 10:21:41 durpdbc2 kernel: Key=4h (HARDWARE ERROR); FRU=03h
Feb 22 10:21:41 durpdbc2 kernel: ASC/ASCQ=40h/01h ""
Feb 22 10:21:41 durpdbc2 kernel: CDB: 28 00 00 00 00 69 00 00 24 00
Feb 22 10:21:41 durpdbc2 kernel:
Feb 22 10:21:41 durpdbc2 kernel: SCSI disk error : host 1 channel 0 id 0 lun 0 return code = 28000002
---< snip >---
These errors appear on both nodes for all disks regularly.
We have a similar cluster up and running which uses exactly the same hardware for the server machines but a different external scsi storage (simple "box" containing 6 x Fujitsu MAS3367NP without backplane).
Both clusters are configured the same way (secondary cluster option enabled in SCSI BIOS, different SCSI IDs for secondary controller). May assumption is that the problem is based in the external SCSI storage backplane as this seems to be the only difference.
Maybe anyone has made the same or a similar experience?
Regards,
Peter
Peter
No Events found!
pfuchs
5 Posts
0
February 28th, 2005 17:00
Hi again,
just in case somebody hits the same problem...
Today I was told by Dell Support that this setup would never work as the LSI controllers would not support clustering. While wondering why it works within the two other servers (most probably because of the missing backplane), I was told that I need to buy a PERC 4/DC controller for each system to make use of the PV220S.
This would have been ok for me if there was some hint on the ordering web site of the PV220S, but there is nothing (at least not on the German one).
Will now look for a good price for the PERCs or give the PV220S back to buy a (much cheaper) SCSI box without backplane.
Best regards,
Peter