2018-12-04 15:21:26 UTC
I've got a distributed replicated 2+1 (arbiter) volume with sharding
enabled, running 3.8.8, for VM hosting, and I need to expand it before I
leave over the holiday break.
Each server's brick is mounted on its own LV, so my plan is the following
with each server, one-by-one:
1. Take the peer offline
2. Add new disks & expand the brick's LV
3. Bring the peer back up
4. Heal the volume, before continuing to the next peer
My question is the following: I know there are reports of the data on
sharded volumes getting corrupted when expanding those volumes. Is that
only the case when expanding a volume by adding new bricks, or can that
also happen if the bricks' LVs are expanded, without making any changes to
the underlying layout of the volume's bricks?
Upgrading the Gluster version isn't an option for the moment, that will
have to wait until next year when I upgrade the entire environment.