On Mon, Aug 9, 2021 at 1:56 PM Gilboa Davara <gilboad@gmail.com> wrote:
>
>
>
> On Mon, Aug 9, 2021 at 12:16 PM Gilboa Davara <gilboad@gmail.com> wrote:
>>
>> Hello,
>>
>> On Mon, Aug 9, 2021 at 11:50 AM Yedidyah Bar David <didi@redhat.com> wrote:
>>>
>>> On Mon, Aug 9, 2021 at 11:43 AM Strahil Nikolov via Users
>>> <users@ovirt.org> wrote:
>>> >
>>> > Corrupted metadata is the problem you see.
>>> >
>>> > I think there was a command to fix it, but I can't recall it right now.
>>>
>>> I think you refer to 'hosted-engine --clean_metadata'. Gilboa - I
>>> suggest to search the net/archives for docs/mentions/discussions of
>>> this option - it's rather drastic. Good luck.
>>>
>>> That said, I must say that if your metadata is corrupted, I wonder
>>> what else is - so would continue using this setup with great care.
>>> Ideally restore from backups, after testing/replacing the hardware.
>>>
>>> Best regards,
>>>
>>
>> Thanks for the pointer.
>> This is a side setup that's about to replaced by a real setup (3 host Gluster).
>> That said, beyond the corrupted meta data, everything else seems to be working just fine, host boot just fine, RAID sync showed no issues. XFS partitions mounted OK, etc.
>> The only thing that seems damaged is the hosted engine meta data.
>>
>> I'll test it and report back.
>>
>> - Gilboa
>>
>
>
> Stupid question:
> Won't clean meta data remove the host from the "cluster" and given the fact that its a single host configuration, require a clean redploy?

It's not stupid.

Generally speaking, the metadata is populated by the HA daemons
themselves, not something "external". If a specific host's entry is
missing, they should write it.