16. October 2024 - verfasst von Oliver Gaida - Kategorien: ["linux", "lvm"]
linux - lvm snapshot before critical changes
make most sense on physical machines, because on virtual machines in most cases it is easier to create a snapshot with the hypervisor toolset.
generell requirements:
- enough space left on the the corrsponding volumegroup
how it goes
my situation:
root@nbsy:~# lvs
LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert
lv-sys vg0 -wi-ao---- 20.00g
test vg0 -wi-a----- 1.00g
root@nbsy:~# df -h
Filesystem Size Used Avail Use% Mounted on
tmpfs 1.6G 1.7M 1.6G 1% /run
/dev/mapper/vg0-lv--sys 20G 7.2G 12G 39% /
tmpfs 7.8G 0 7.8G 0% /dev/shm
tmpfs 5.0M 0 5.0M 0% /run/lock
/dev/sda1 1.1G 6.1M 1.1G 1% /boot/efi
/dev/sda3 590G 28K 560G 1% /data
tmpfs 1.6G 4.0K 1.6G 1% /run/user/1000
root@nbsy:~#
create snapshot
lvcreate -s -L 10G -n snap_initial /dev/vg0/lv-sys
do not keep changes / delete everything and go back to the snapshot version
# back to before-state:
lvconvert --merge /dev/vg0/snap_initial
# reboot the system
# refresh lvstate:
lvchange --refresh vg0
# sometimes it need some time, check with:
lvs -a
# create again the original snapshot, because after merging it was deleted. But wait till the merge of the old snapshot has been finished
lvcreate -s -L 10G -n snap_initial /dev/vg0/lv-sys
keep changes and do more testing
lvremove /dev/vg0/snap_initial
lvcreate -s -L 10G -n snap_initial /dev/vg0/lv-sys