lifls

[Steve Bear_1]

I did "lifls /dev/rdsk/cxtxdx" to see the lif area but suddenly after a while i saw the following messages in the syslog.log

Any idea ? why ? what was wrong with that ?


------
Oct 28 08:52:51 test_nts vmunix: LVM: Performed a switch for Lun ID = 0 (pv = 0x0cf76400), from raw device 0x1c040000 (with priority: 0, and current flags: 0x240) to raw device 0x1c011000 (with priority: 1, and current flags: 0x0).
Oct 28 08:52:51 test_nts vmunix: LVM: Performed a switch for Lun ID = 0 (pv = 0x0cf76400), from raw device 0x1c040000 (with priority: 0, and current flags: 0x240) to raw device 0x1c011000 (with priority: 1, and current flags: 0x0).
Oct 28 08:52:51 test_nts vmunix: LVM: vg[2]: pvnum=0 (dev_t=0x1c011000) is POWERFAILED
Oct 28 08:52:51 test_nts vmunix: LVM: vg[2]: pvnum=0 (dev_t=0x1c011000) is POWERFAILED
Oct 28 08:52:51 test_nts vmunix: LVM: PV 0 has been returned to vg[2].
Oct 28 08:52:51 test_nts vmunix: LVM: PV 0 has been returned to vg[2].
Oct 28 08:52:51 test_nts vmunix: LVM: A link to PV 0 in vg[2] has been restored.
Oct 28 08:52:51 test_nts vmunix: LVM: A link to PV 0 in vg[2] has been restored.
Oct 28 08:52:51 test_nts vmunix: LVM: Performed a switch for Lun ID = 0 (pv = 0x0cf76400), from raw device 0x1c011000 (with priority: 1, and current flags: 0x280) to raw device 0x1c040000 (with priority: 0, and current flags: 0x80).
Oct 28 08:52:51 test_nts vmunix: LVM: Performed a switch for Lun ID = 0 (pv = 0x0cf76400), from raw device 0x1c011000 (with priority: 1, and current flags: 0x280) to raw device 0x1c040000 (with priority: 0, and current flags: 0x80).
----

[Stefan Farrelly · Accepted solution]

One of your disks powered down ("POWERFAILED"), and as you had alternative paths LVM switched over to the other disk ("Performed a switch" - preusmably the mirror) but in time the disk which powered down came back up ("returned to VG").

so everything is ok now, but the disk which powered down is possibly dying. Still, as your mirrored nothing to worry about for now.
I would try your lifls command again to see if causes the same problem, if so then I would log a hardware call with HP and get it replaced - before it dies completely.

[Steve Bear_1]

Thanks Stephen,

Yes, error messages written to syslog just exactly after i typed "lifls"

[Zeev Schultz]

lifls shouldn't cause PV link change (thats what happened due to syslog).It jumped from cXt0d0 to cYt1d0. To see X and Y - ioscan -fnCext_bus (look for instance numbers 4 and 1).See if the drive you did lifls to is one of those.Check it also with dd if=/dev/rdsk/cXtYd0 of=/dev/null bs=1024k.

[Steve Bear_1]

yes, i can confirm it was one of those, i did lifls on /dev/rdsk/c1t1d0 (raw device 0x1c011000 )

[Zeev Schultz]

Yes but it jumped TO /dev/rdsk/c1t1d0
which wasn't active that time (c0t4d0 was).
What LVM did is to make c1t1d0 active for a while (time that took to lifls complete) and jump back to c0t4d0.I can guess despite that lifls doesn't touch LVM (works on raw device) -
LVM acts as it does. I'd check both drives (basic check with dd), see no error or scsi messages were there. If no other error messages or failures are there - monitor the drive for a while and let it be.