volume replication

[jmo]

Hi butre,

first of all: I have no practical experience with open-e DSS iSCSI replication.
second: As far as I known DSS is using DRBD for replication.

[...]
1) when setting up replication of the volumes of 10gb the initial replication replicates the full 50gb (according to the task status) although it contains no data. is this expected behaviour?

Yes, that's the behaviour I'd expect: the *device* is replicated/synchronized, not the data.

It like with RAID syncs: when a device fails, all blocks need to be resynced, not only those containing data. DRBD doesn't know which blocks are empty, which are full - only the file system knows.

2nd question has to do with additional replication after initial replication.

• when i add the volume to my xenserver and replicate it replicates about 100mb of additional data
• when i put a 8gb virtual machine on the volume (with 3.5gb of data and replicate it only replicates the 3.5gb (which is good)

• [As thise are the blocks changed on the (replication) device, only those blocks are sent.
  
  

• i then stop volume replication, make the destination volume a source, mount it in xenserver and i can access the virtual machine, after making some slight tweaks i run the reverse replication job (on the primary i toggle source to destination) but to my surprise it replicates the full 10gb volume and not just the changes.