[linux-lvm] LVM cachepool inconsistency after power event

Ming Hung Tsai mtsai at redhat.com
Fri Oct 8 07:56:48 UTC 2021

On Thu, Oct 7, 2021 at 3:57 PM Krzysztof Chojnowski <frirajder at gmail.com> wrote:
> On Tue, Oct 5, 2021 at 6:57 PM Ming Hung Tsai <mtsai at redhat.com> wrote:
>> On Tue, Oct 5, 2021 at 11:54 PM Zdenek Kabelac <zkabelac at redhat.com> wrote:
>> > lvconvert --repair should be able to handle this case - although likely
>> > without 'smart' placement' of fixed metadata (pvmove needed after metadata fix)
> I tried running lvconvert --repair earlier (please see the 1st of my mails) and even trough the operation finished successfully (as in the $? == 0) I couldn't activate the volume afterwards.
>> cache_repair might not handle this case perfectly. If possible, you
>> could upload the "original" metadata somewhere for me to take a look,
>> by dd it into a binary file.
> Unfortunately I deleted the original metadata volume when I was trying to remove the cached volume.
> So right now, I guess, the question is how do I remove the cached volume, as the repair doesn't seem to be possible.


You could simply detach the cache:
sudo lvconvert vg0/tpg1-wdata --splitcache

Or alternatively, reinitialize the cache by swapping the metadata with
a blank one, which bypasses cache_check examination and then steps to
metadata formatting:
sudo lvcreate vg0 --size 40m --name cmeta_bak
sudo lvconvert vg0/wdata_cachepool_cpool --swapmetadata --poolmetadata cmeta_bak

then you should be able to activate the tpg1-wdata.

BTW, could you help provide the reproducing steps? I guess that you
performed a power cut soon after creating the cache. Thanks.

Ming-Hung Tsai

More information about the linux-lvm mailing list