Issue

• We are making heavy usage of mirrored logical volumes to implement host based replication of SAN storage.
• If one of the two mirror legs goes absent and comes back after a certain time, we have to do a full resync.
• With growing logical volume sizes this takes way too long to ensure SLAs and during that time, we are prone to lose the second leg as well and will be left with an unusable system.
• Set up a mirrored logical volume, remove one leg, write data to the second then re-add the first leg. If it does not need a full sync, but just a sync of the changed data, this requirement would be fulfilled.
• This RFE currently already exists upstream via http://www.sourceware.org/lvm2/wiki/FeatureRequests/dm/mirroring.