Currently, to support disaster recovery for RBD, we advise that customers export/import each RBD as an image for copying between two distinct clusters. This is cumbersome with large numbers, requires a dedicated host to manage the files and can be slow for the initial copy.
Async RBD Mirroring will allow each RBD to be streamed in realtime to a secondary cluster and be point in time consistent. This feature is roughly equivalent to NetApp's SnapMirror in purpose.
* As an administrator, I select a pool's RBD contents to be asynchronously mirrored to a pool in a second RADOS cluster.
* As an administrator, I select an individual RBD to be asynchronously mirrored to a pool in a second RADOS cluster.
* As an administrator, I can see which pools or RBDs are successfully being mirrored to a secondary cluster.
* As an administrator, I am warned when a pool or RBD is not being successfully mirrored to a secondary cluster.
* As an administrator, I can pause mirroring for a pool or RBD.
There will be a new daemon "rbd-mirror" that runs on a separate server, similar to how we isolate RGW on its own node(s). It is essentially a client to the cluster, so same constraints as RGW.
Looks good to me -- only minor comment is perhaps mention that it is asynchronous and crash consistent mirroring?
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.