topic Re: dmesg in Operating System - HP-UX

dmesg

Marcos Tinoco Garcia — Wed, 09 Oct 2002 15:01:04 GMT

Hi! I'm Marcos and ...
When I run "dmesg" command in hp9000-cpnv1 with hpux-11.00, I see a message that it show this.

"LVM: Recovered Path (device 0x1f06a300) to PV 1 in VG 5.
LVM: Performed a switch for Lun ID = 0 (pv = 0x0000000050249000), from raw devic
e 0x1f07a300 (with priority: 1, and current flags: 0x40) to raw device 0x1f06a30
0 (with priority: 0, and current flags: 0x0).
LVM: Restored PV 1 to VG 5.
LVM: vg[5]: pvnum=1 (dev_t=0x1f06a300) is POWERFAILED
LVM: Recovered Path (device 0x1f07a300) to PV 1 in VG 5."

Hi! I'm Marcos and ...
When I run "dmesg" command in hp9000-cpnv1 with hpux-11.00, I see a messages that it show this.
How do I relate it message with the disk configured in LVM ?
What Does "pvnum=1" mean ?
What Does "device 0x1f07a300) to PV 1 in VG 5" mean ?
... etc
Thank's

Re: dmesg

Cheryl Griffin — Wed, 09 Oct 2002 15:07:37 GMT

"pvnum=1" is the first physical volume.
"device 0x1f07a300" is the device in hex reporting the problem.
# ll /dev/dsk |grep 07a300
07a300 = the minor number
0x1f = the major number in hex = 31
# lsdev |grep disk
"vg 5" is the 5th volume group; vg00 = 1, vg01 = 2 .... vg04 = 5

# strings /etc/lvmtab
Make sure that you have PHKL_18543 (major SCSI/LVM patch) installed, configured.

You may be having a disk go bad. Look for errors in /var/adm/syslog/syslog.log.

Cheryl

Re: dmesg

John Poff — Wed, 09 Oct 2002 15:11:21 GMT

Hi Marcos,

It looks like you have a PV or a controller going bad. Here is how to decode the dev_t info and figure out which disk it is. The 1f is the major number of the disk device, which is 31. The 06 is the card instance number. The a is the target, 10, and the 3 is the LUN. This looks like c6t10d3 to me. Likewise, the 0x1f07a300 would be disk c7t10d3.

JP

Re: dmesg

S.K. Chan — Wed, 09 Oct 2002 15:17:20 GMT

In addition to others ..the disk c6t10d3 in vg05 (which is the 1st PV in vg05) looks to me that it may have some problem related to ..
- Low timeout value (this can be changed with "pvchange -t")
- Possible IO or disk problem
- Possible SCSI termination or connection problem.
I would first run diagnostics on the suspected disk to try eliminate all the above possibilities one by one. How frequent are you getting this error in syslog.log ?

Re: dmesg

Jeff Schussele — Wed, 09 Oct 2002 16:06:19 GMT

Hi Marcos,

I agree with S.K.
A lot of link switches can be traced to low timeout values. We set ours up to 180 seconds & when we switch we almost always have some sort of "real" problem.
Check the timeout value on the LV with the lvdisplay command
lvdisplay /dev/vg_name/lv_name

Rgds,
Jeff