Feature added by: Lars Marowsky-Bree (LarsMB) Feature #311123, revision 1 Title: Support for replicated root filesystem (DRBD) Hackweek VI: Unconfirmed Priority Requester: Important Requested by: Lars Marowsky-Bree (larsmb) Partner organization: openSUSE.org Description: The idea is to support the root file system (or rather, the root block device) on top of DRBD (http://www.drbd.org/). There are at least two use cases. The first is for a VM guest: it allows the guest to be replicated (or have a fully uptodate backup image) somewhere even on a cloud where the hypervisor does not support the replication of images. The second is for a physical host: this allows an uptodate root image of the node. This can be used to recover the node in case of a lost local disk, as a backup, or for P2V/P2P migration. (Possibly by bringing the protected work load up on a virtual host immediately once the hardware has failed, or on another node.) The benefit of using drbd is that the node is immediately usable, without waiting for the resync to complete - DRBD will redirect reads that can't (yet) be completed locally to the backup copy. It is immediately usable. The remote side would be a Linux host running DRBD as well, of course ("replica master"). A single replica master can host any number of DRBD end points. It seems feasible to have this replica master be a Studio Appliance, or even use a pacemaker cluster that manages the end points; this side would always only be a "slave" instance. DRBD can also cope well with temporary connection loss or the replica master rebooting; once restored, it will resync only the since modified data. A further use case is that snapshots/backups of the workload can be taken on the replica master without impacting the live workload. -- openSUSE Feature: https://features.opensuse.org/311123