"LVM failed to initialize library context" - SLES 12.3 after reboot lost some of its PVs

Question

After a reboot, a SLES 12.3 box doesn't recognizes a few disks (that are actually behind HW raid).

What could the issue be?

If I "head /dev/sdb | strings | head" the problematic disks, I can see data on them, ok.

But "pvscan" doesn't recognizes it.

SERVER:~ # pvck /dev/sdb
  Device /dev/sdb not found (or ignored by filtering).
SERVER:~ # 
SERVER:~ # grep '  filter' /etc/lvm/lvm.conf
    filter = [ "r|/dev/.*/by-path/.*|", "r|/dev/.*/by-id/.*|", "r|/dev/fd.*|", "r|/dev/cdrom|",  "a|/dev/sd.*[1-9]|", "a|/dev/vda[1-9]|", "a|/dev/mapper/.*|", "a|/dev/dm-*|" , "r|.*|" ]
SERVER:~ #

Strangely, the OS recognizes sda, which holds the OS itself.

But it doesn't recognizes the sdb/sdc/etc (holding the data) as PV.

UPDATE on 2018 Dec 11: all the disks: sda (raid1), sdb(raid5), sdc(raid5), sdd(raid1) are on HW RAID. "zypper up" says nothing to update.

There is a know bug after upgrades, especially when using multipath. — Rui F Ribeiro, Dec 10 '18 at 20:17
hmm, no multipath is used here, only HW RAID1 (which is sda) and HW RAID5, which is RAID5. But many thanks for any hint. — KollarA, Dec 11 '18 at 09:48
Yes, but probably does not apply. https://www.suse.com/support/kb/doc/?id=7022520 — Rui F Ribeiro, Dec 11 '18 at 11:21
sda is from HW RAID too and that works, but checking, Thanks! — KollarA, Dec 11 '18 at 11:24

score 0 · Answer 1 · answered Dec 11 '18 at 12:52

finally, we found out:

It looks like something deleted the non-sda partitions :)

So the data was there on the ex.: sdb, sdc, sdd, but we had to re-run the partition creating oneliner:

parted -s /dev/sdb mklabel gpt mkpart primary xfs 0% 100% set 1 lvm on

and presto, pvscan found them! yay.

"LVM failed to initialize library context" - SLES 12.3 after reboot lost some of its PVs

1 Answers1