No announcement yet.

DRBD support (comeback or old feature)

  • Filter
  • Time
  • Show
Clear All
new posts

  • DRBD support (comeback or old feature)

    Would it be possible to "officially" support DRBD again? You had it at the beginning but then removed if when moving to ZFS backend...

    It would be fantastic to be able to have a "shared nothing" cluster by (live) replicating a storage pool via DRBD between two machines.
    DRBD evolved quite a but from back then... and it would be very useful.

    It's somehow working with manual tweaking but having it integrated in the GUI/CLI would be excellent!

    Keep the good work with this excellent product!

  • #2
    Thanks Mauricio,
    We don't have any current plans to re-integrate DRBD as our road-map is set for the next couple quarters but I do think it's a good idea to review DRBD again in the future. As you note, we did some integration work with it back at version 8.3 and I know they've been actively improving it over the years. One of the things needed to replicate a pool is the ability to atomically replicate a group of block devices as a consistency group. It looks like they have a concept like that in DRBD9 where volumes can share a replication stream but I'm not sure if that replication logic is atomically updating all volumes to a common point-in-time in lock step on each update. Without something like that a given zpool could easily become corrupted on the target/receiver side and not startup properly on failover.