>>>"Try to get all data in advance (before deactivating
the VG)".
Can you clarify? What do you mean by this?
Get all necessary info before disabling the VG.
For example:
pvs
vgs
lvs -a
lvdisplay -m /dev/gluster_vg1/lvthinpool
lvdisplay -m /dev/gluster_vg1/lvthinpool_tmeta
lvdisplay -m /dev/gluster_vg1/lvthinpool_tdata
>>>"You got 15GiB of metadata, so create your new
metadata LV at least 30 GiB."
My presumption is that 15 GB is the max size the was specified when it was initially
built but not the actual size right now, but yes, making it larger does make sense. I was
curious to know what the present size is. Also, lvconvert man page specifies a supported
value between 2M and 16G I will create a larger metadata volume assuming I can get the
procedure to work properly I am having some difficulty with the procedure, see below.
If you don't manually create the meta and data LVs (which will create the pool with
another command) the size is created by the following formula:
(Pool_LV_size / Pool_LV_chunk_size * 64).
For details check:
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/...
My most important question was here:
I am trying to follow the procedure from
https://access.redhat.com/solutions/3251681 I am
on step #2. Step 2a and 2b work without issue. Step 2c gives me an error.
Here are the values I am using:
# lvcreate -L 2G gluster_vg3 --name tmpLV
(created with no issues)
# lvchange -ay gluster_vg3/tmpLV
(command completed with no issues)
# lvconvert --thinpool gluster_vg1/lvthinpool --poolmetadata gluster_vg3/tmpLV
Error:
VG name mismatch from position arg (gluster_vg1) and option arg (gluster_vg3)
Do I need to create the LV on the same disk that failed? (gluster_vg1)
According to the error yes - logic shows that you need to be working in the same VG.If
you don't have free space on the thin pool, you can add additional PV and extend the
pool and then create your thin meta LV.
You can also merge the 2 VGs to get the necessary space. Check 'man 8 vgmerge'.
Best Regards,
Strahil Nikolov