topic Re: SCSI: Parity Error - recent interrupt: in Operating System - HP-UX

SCSI: Parity Error - recent interrupt:

LITWINCZUK — Thu, 01 Sep 2011 10:16:51 GMT

How to recognize what device have this problem? (and why)

Found in syslog:

Aug 30 15:31:52  vmunix: SCSI: Parity Error -- lbolt: 397763214, dev: cb058002
Aug 30 15:31:52  vmunix:            lbp->state: 2060
Aug 30 15:31:52  vmunix:            lbp->offset: ffffffff
Aug 30 15:31:52  vmunix:            lbp->uPhysScript: fbfef000
Aug 30 15:31:52  vmunix:    From most recent interrupt:
Aug 30 15:31:52  vmunix:            ISTAT: 0a, SIST0: 11, SIST1: 00, DSTAT: 80, DSPS: 00478300
Aug 30 15:31:52  vmunix:    lsp: 0000000065cea400
Aug 30 15:31:52  vmunix:            bp->b_dev: cb058002
Aug 30 15:31:52  vmunix:            scb->io_id: 5006b66
Aug 30 15:31:52  vmunix:            scb->cdb: 12 00 00 00 80 00
Aug 30 15:31:52  vmunix:            lbolt_at_timeout: 0, lbolt_at_start: 0
Aug 30 15:31:52  vmunix:            lsp->state: 5
Aug 30 15:31:52  vmunix:    lbp->owner: 0000000000000000
Aug 30 15:31:52  vmunix:    scratch_lsp: 0000000065cea400
Aug 30 15:31:52  vmunix:    Script dump [0000000058e6e000]:
Aug 30 15:31:52  vmunix:            09000080 00478300 e25c0004 fbfef7f8
Aug 30 15:31:52  vmunix:            80080000 fbfef090 e25c0004 fbfef7f8
Aug 30 15:31:55  vmunix: SCSI: Resetting SCSI -- lbolt: 397763314, bus: 5
Aug 30 15:31:55  vmunix: SCSI: Reset detected -- lbolt: 397763314, bus: 5

Re: SCSI: Parity Error - recent interrupt:

Matti_Kurkela — Thu, 01 Sep 2011 12:48:02 GMT

 dev: cb058002

This is the major & minor device number encoded as a single value in hexadecimal.

The first byte "cb" is the major device number. To convert it to decimal, run:

$ printf "%d\n" 0xcb
203

The major number identifies the type of the driver. You can find the numbers in "lsdev" listing. That seems to refer to "sctl" devices. The "scsi_ctl" man page describes how to decode the minor number for the sctl devices. This is SCSI pass-through driver, which is often used for controlling the robotics of a tape library, or for other special purposes.

The minor number is 058002: this decodes to SCSI controller #5, target #8, LUN #0 and optional value 2 (no Inquiry on open). The standard name for such a device would be /dev/rscsi/c5t8d0, but as the SCSI pass-through devices are created by the sysadmin on a case-by-case basis, a different name might have been used.

To find the name of the actual device node, run this command:

find /dev -type c | xargs ll | grep "203 0x058002"

To find the hardware path of the SCSI controller #5, you might do this:

ioscan -I 5 -C ext_bus -fk

The next step would be to physically check the cable attached to this SCSI controller (the hardware path of each card slot is typically listed in server documentation and/or printed next to the card slot). Inspect all the devices connected to this controller, and find the one that has its SCSI ID set to #8.

The problem was a parity error in the SCSI communication, so it might be caused by a damaged cable or a bad connection. Make sure all the connectors are fully plugged in and none of the cables are kinked or otherwise damaged.

Re: SCSI: Parity Error - recent interrupt:

Viktor Balogh — Thu, 01 Sep 2011 13:01:46 GMT

cb058002 refers to the major and minor number of the device. The format is:

MMBBTLFF where

MM = Major number
BB = Bus number
T = Target
L = LUN
FF = Flags

cb is hex, by converting it to decimal it suggests that the major number of the device in question is 203.

Now do this to find the device:

# find /dev -exec ls -lrtd {} \+ | awk '$5 == 203 && $6 ~ /058002/ '

Regards,

Viktor

Re: SCSI: Parity Error - recent interrupt:

LITWINCZUK — Thu, 01 Sep 2011 18:09:03 GMT

Thanks for full explain.