Closed resdigita closed 4 years ago
Hmn, that's unfortunate. I'm also using LXD with lvm2. However, I've simply updated the package so I didn't notice yet this issue yet.
According an ongoing discussion there might be a solution in future releases.
Do you have any link about that? I'd be interested.
I just re-built the EPEL-8 packages for lxd-3.18-0.2
. You can manually download them from: https://linuxmonk.ch/packages/lxc3/epel/8/
Unfortunately COPR only keeps the latest successful built. I'm trying to setup a COPR package archive, so that old packages continue to be available for such cases in the future.
As long as you just update an existing installation you won't notice. I updated from CentOS 7 to 8 which is a new installation and got stuck. The discussion is here: (https://discuss.linuxcontainers.org/t/error-when-trying-to-use-an-existing-lvm-pool-with-a-new-installation/6809)
And thank you very much for version 3.18! I just downloaded everything. Thankfully I don't have to wait for the new version.
Many thanks!
Just finished the reconstruction of the LXD runtime on two of our servers including restore of the most important containers using the 3.18 version you provided. Got a little late (or very early, depending on perspective), but I'm glad that it all worked out perfectly.
Again, many thanks!
@resdigita - how did you downgrade? I am getting the following:
Error: Error creating database: schema version '39' is more recent than expected '38'
after downgrading on my system from 3.20 -> 3.18.
@mrmateo - I didn't downgrade an existing and working installation. I upgraded from CentOS 7 to 8, so I had to perform a completely new installation. I use a procedure to perform a backup of each container. In CentOS 8 I created a new lvm pool and restored each container from that backup.
Good luck!
how did you downgrade? I am getting the following:
Error: Error creating database: schema version '39' is more recent than expected '38'
after downgrading on my system from 3.20 -> 3.18.
I think downgrading is "officially" not supported. However LXD is creating a backup of the database before each upgrade. So if you want to downgrade you can try to copy back the /var/lib/lxd/database/local.db.bak
to /var/lib/lxd/database/local.db
.
Thanks for the info! I will remember that if I need to do it in the future.
I ended up fully removing the sub-volumes and then doing a fresh install and that worked. Had to re-create my containers but that didn't take too long. I guess it's a learning experience to test the upgrade first before doing it on the system where I'm actually using it :)
I few days ago, I built lxd-3.22-0.1
that should restore the LVM support on an VG with existing LVs.
lxd/storage/drivers/driver/lvm: Adds lvm.vg.force_reuse config option
There is still a build problem with the >=Fedora 32 packages but since you were using CentOS I'm going close this issue.
With version 3.20 it is no longer possible to create an LVM pool in a VG where LVs already reside (which should be the common case). According an ongoing discussion there might be a solution in future releases. In the meantime I have to go back to version 3.18 where this was still possible (considered a bug by developers).
Are there 3.18 (or earlier) packages for Epel 8 available anywhere? Unfortunately, I could only find 3.20 rpms so far.