LUN Switch Errors in Dmesg Oracle 9i

[Saurav_1]

Hello Every Body,

We have connected a new HP box rp4440,HP-UX 11.11 to a VA7410. The setup was quite stable before that. Few LUNs are connected to the HP-UX box. We are running 9 databases on this server of Oracle 9.2. Glance shows the Filesystem Utilization 100% most of the time. Due to this some Some Luns go offline and following error messgae comes. Switch to alternate link.
______________________________________________
May 19 01:49:26 ebe vmunix: SCSI: Read error -- dev: b 31 0x097200, errno: 126, resid: 2048,
May 19 01:49:26 ebe vmunix: blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
May 19 01:53:58 ebe vmunix: LVM: Recovered Path (device 0x1f097200) to PV 0 in VG 18.
May 19 01:49:26 ebe vmunix:
May 19 01:53:58 ebe vmunix: LVM: Restored PV 0 to VG 18.
May 19 01:55:56 ebe vmunix: LVM: Recovered Path (device 0x1f097200) to PV 0 in VG 18.
May 19 01:55:56 ebe vmunix: LVM: Restored PV 0 to VG 18
_______________________________________________

The PV timeout value is changed to 180 Seconds on this server. THese messages are common to other servers also running HP-UX 11i and 11.

Following patches have been installed to resolve this issue but all in vane.
PHKL_32090 SCSI I/O PATCH

PHKL_32089 SCSI I/O CUMULATIVE PATCH

PHCO_30698 LVM CUMULATIVE PATCH (addresses FC aborts/Corruption)

PHKL_31216 LVM CUMULATIVE PATCH

PHKL_30257 FIBRE CHANNEL DRIVER UPDATE.

Max points assured to all the replies and help.

Thanks in advance.

[Eric Antunes · Accepted solution]

Hi Saurav,

Do you have PHKL_31501 (or any equivalent for your HP-UX version) for Fiber Channel Mass Storage installed?

Best Regards,

Eric Antunes

[Naveej.K.A]

Hi,

Do you actually have any faults on the VA7410 controllers?? Are the array's LED indications all showing "good"??

Any hard error reported through STM??

Regards,
Naveej

[Saurav_1]

Thanks for the reply. But the respective patch for HP-UX11.11 is already installed. However I receive the following error in the array log.
______________________________________
Event Description = FRONTEND_FC_ABTS_EVENT_EH This error code indicates that the Host sent a Fibre Channel ABTS (Abort Sequence) BLS frame to the abort an IO. The array will log this event for informational and debug purposes only. It does not necessarily indicate a problem with the array.
_____________________________________

[Devender Khatana]

Hi,

In concern to the messages apearing I would suspect something like cables as well.On our SCSI devices we had this problem some time back and was found to be a SCSI cable problem. You are accessing your LUNs through multiple paths there is no way in LVM to show currently available paths like "vxdg listctlr all" in Veritas. The only way to verify this is to check the LUNs using diskinfo command at the time when the error has occurred and has path has not been restored.

HTH,
Devender

[Saurav_1]

Thanks Devender.

I have found some FC aborts in Array logs and found a forum topic which matches my problem itself. It says to change the resilency setting to default.