[linux-lvm] " Failed to find physical volume <ZFS block device"

Roger Heflin rogerheflin at gmail.com
Tue Sep 28 18:47:06 UTC 2021

A possibility I just debugged for a non-booting system.

If there is a partition table on the underlying device then that
device is not detected as an LVM1/2 member in at least one version of
udevd, and won't be seen nor turned on automatically by the
systemd-udev code.

lvm vgchange -ay worked to enable it (emergency mode, it was the root
pv--no udevd involvement) and eventually I found the partition table
and removed it and the machine would then boot without needing a
manual intervention.

dd if=/dev/zero of=device bs=512 count=1 was used once we determined
there was a partition signature still left (after partition deletion
with fdisk, still had a header), examined with dd if=/dev/device
bs=512 count=1 | xxd found 4 non-zero bytes in the block.

On Sun, Sep 26, 2021 at 9:22 AM alessandro macuz
<alessandro.macuz at gmail.com> wrote:
> Thanks Roger, Zdenek,
> I have my ZVOL on my NAS exposed as LUNs. The initiator were switched off and for unknown reason I found my NAS switched off as well.
> It had run for long and I feared the worst (CPU/motherboard/etc). Instead once powered up everything started to work again but the LUNs that seemed to jeopardized.
> I have many ZVOLs used by ESXi in which I have EVE-NG who uses LVM and such ZVOLs have the same size so I wanted to inspect them to check the hostname.
> Now some LUNs started to work normally, some others still behave weirdly. I will run pvs on them with extra debugs to see what's going on.
> Many thanks,
> Alex
> Le jeu. 23 sept. 2021 à 23:48, Roger Heflin <rogerheflin at gmail.com> a écrit :
>> If you have lvmetad running and in use then the lvm commands ask it
>> what the system has on it.
>> I have seen on random boots fairly separated systems (rhel7 versions,
>> and many years newer fedora systems) at random fail to find one or
>> more pv.s
>> I have disabled it at home, and in my day job we have also disabled
>> (across 20k+ systems) as we confirmed it had inconsistency issues
>> several times on a variety of our newest installs.
>> Stopping lvmetad and/or restarting it would generally fix it.    But
>> it was a source of enough random issues(often failure to mount on a
>> boot, so often issues that resulted in page-outs to debug)  and did
>> not speed things up much enough to be worth it even on devices with
>> >2000 SAN volumes.
>> On Thu, Sep 23, 2021 at 8:52 AM Zdenek Kabelac <zkabelac at redhat.com> wrote:
>> >
>> > Dne 22. 09. 21 v 18:48 alessandro macuz napsal(a):
>> > > fdisk correctly identifies the extended partition as 8e.
>> > > I wonder which kind of data lvmdiskscan and pvs use in order to list LVM
>> > > physical volumes.
>> > > Does PVS check some specific metadata within the partition without just
>> > > relying on the type of partition displayed by fdisk?
>> > >
>> > >
>> >
>> > Hi
>> >
>> > Yes - PVs do have header signature keeping information about PV attributes
>> > and also has the storage area to keep lvm2 metadata.
>> >
>> > Partition flags known to fdisk are irrelevant.
>> >
>> >
>> > Regards
>> >
>> > Zdenek
>> >
>> > _______________________________________________
>> > linux-lvm mailing list
>> > linux-lvm at redhat.com
>> > https://listman.redhat.com/mailman/listinfo/linux-lvm
>> > read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
>> >

More information about the linux-lvm mailing list