Issues with Make_tape_recovery & LIF

[Torsten.]

If the disk is UNAVAILABLE in LVM already, you can pull the disk.

If not, you need to do a "pvchange -a n ..." before.

But after inserting the new disk you probably need to do a

# scsimgr replace_wwid –D /dev/rdisk/disk100

just do an ioscan after insertion and watch the syslog for related messages.

[EU-Admins-UNIX]

Strange things are happening Torsten.

Before Eng. went away to get a new disk, he put in the failing one again just to prove it was CLAIMED, andf it was.  Now, it has become available in vgdisplay, no stale extents, but lvlnboot -v looks strange.  see attachment:

[Torsten.]

It looks good!

Test with "dd" again for errors.

The "strange" part is here:

lvlnboot: LIF information corrupt or not present on "/dev/disk/disk87".


This is a data disk with a boot area, but the area is empty.


You can use "vgmodify" to remove or just ignore.

[EU-Admins-UNIX]

dd failed:

root@apollo at /var/adm/syslog =>dd if=/dev/rdisk/disk100 of=/dev/null bs=1024k count=64
dd read error: I/O error
1+0 records in
1+0 records out

[Torsten.]

So the disk is probably "half-dead".


AFAIR the newest LVM version has this "half-dead-detection", so it will disable the LVM access - this could be the reason why it was unavailable.


You should replace the disk.

[EU-Admins-UNIX]

Should I

lvreduce -m 0 -A n /dev/vgoo/lvname /dev/disk100 (for each LV on that PV)? 

I'm not 100% sure whether this is necessary or not?

[Torsten.]

Don't reduce anything!

Just disable LVM access (follow the good_disk_go_bad document)!



BTW, see also

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c03233055/c03233055.pdf

Half Dead Disk Detection

A physical disk that returns media error for write I/Os while allowing read I/Os to succeed (or
vice-versa) is termed a Half Dead Disk. The term also refers to a disk that returns media error for
certain regions while the rest is unaffected. When a disk is fully dead (i.e. both read and write
I/Os to any region of the disk fail), LVM detects it automatically and marks it as unavailable. But
LVM did not handle half dead disk conditions until this release. The half dead disk situation causes
significant performance impact to the system since LVM indefinitely attempts recovery of the affected
physical volumes. The new Half Dead Disk Detection feature can be used to detect and quickly
exclude such PVs from LVM usage.

[EU-Admins-UNIX]

This get stranger and stranger.

Pulled our failing disk.  Tried 3 different new disks, all showed NO_HW.  'insf -e' did nothing!

Put the original failing disk back in, showed as claimed.

Our disk has firmware version HPC0, engineer brought 3 x disk with fw version HPC3.  They are ordering some HPC0 disks to see if that is the issue.

Put original disk back in until new one arrives.

Thanks for your help.

Will keep you posted.

Tariq

[Torsten.]

I don't think it is strange.

Did you watch the syslog file as I told you???

[EU-Admins-UNIX]

oops