Hi!
I think everything is behaving well but i would still like to ask for confirmation. As starting point there is PVE v. 7 cluster and intent is to upgrade it to v. 8. It will go node by node. I thought to secure my upgrade attempt so
1. i am arriving to next pve node i am going to upgrade
2. i make root filesystem copy to other lvm resource (say original is at /dev/pve/root and i create /dev/pve/root_7)
3. i upgrade node the usual way
4. start using it as v 8
5. after some time say two days i find i did overlook something and i am not happy
6. i boot node up /dev/pve/root_7 being root filesystem
7. i am back to what it was when having this node in cluster as v 7 node
This is something i have done and it works for me well. I also looked at /var/lib/pve-cluster/config.db with sqlitebrowser at different pve versions (6, 7, 8) and it seems very similar in structure.
for the record, this /dev/pve/root_7 is created like this (even if it has mistakes in itself dual booting as dual booting linux operating system works properly)
I would be thankful if you could comment on this booting same node as v. 7 and v. 8 in turns, does it have any negative effect on cluster behavior or cluster configuration. Or there is something to consider i do not know even to be afraid of.
Best regards,
Imre
I think everything is behaving well but i would still like to ask for confirmation. As starting point there is PVE v. 7 cluster and intent is to upgrade it to v. 8. It will go node by node. I thought to secure my upgrade attempt so
1. i am arriving to next pve node i am going to upgrade
2. i make root filesystem copy to other lvm resource (say original is at /dev/pve/root and i create /dev/pve/root_7)
3. i upgrade node the usual way
4. start using it as v 8
5. after some time say two days i find i did overlook something and i am not happy
6. i boot node up /dev/pve/root_7 being root filesystem
7. i am back to what it was when having this node in cluster as v 7 node
This is something i have done and it works for me well. I also looked at /var/lib/pve-cluster/config.db with sqlitebrowser at different pve versions (6, 7, 8) and it seems very similar in structure.
for the record, this /dev/pve/root_7 is created like this (even if it has mistakes in itself dual booting as dual booting linux operating system works properly)
Code:
# lvcreate -L24G -n root_7 pve
# mkfs.ext4 /dev/pve/root_7
# mkdir /mnt/pve-root /mnt/pve-root_7
# mount /dev/pve/root_7 /mnt/pve-root_7
# mount --bind / /mnt/pve-root
# rsync -avHAX --numeric-ids --exclude=/mnt /mnt/pve-root/ /mnt/pve-root_7/
# mkdir /mnt/pve-root_7/mnt
# nano /mnt/pve-root_7/mnt/etc/fstab
# umount /mnt/pve-root_7
# umount /mnt/pve-root
# nano /boot/grub/grub.cfgI would be thankful if you could comment on this booting same node as v. 7 and v. 8 in turns, does it have any negative effect on cluster behavior or cluster configuration. Or there is something to consider i do not know even to be afraid of.
Best regards,
Imre