search

vatesfr / xen-orchestra

The global orchestration solution to manage and backup XCP-ng and XenServer.
https://xen-orchestra.com
Other
767 stars 262 forks source link

Continuous replication does not replicate windows 2012r2 gpt disk correctly #2287

Closed buzzzo closed 6 years ago

buzzzo commented 7 years ago

Source xs: 6.5 Target xs: 7.1 xen orchestra from sources: xo-server + xo-web 5.11

Expected behavior

A windows 2012r2 with 2 disk (c: mbr,p:gpt 100gb) is configured to be replicated on second node

The expected behavior is to: 1) clone the vm 2) start it 3) see data on both disk (c:,p:)

What's really happened: 1) vm correctly boots 2) disk p: is offline 3) bring up via windows disk management 4) disk p: is empty.

olivierlambert commented 7 years ago

Do you have quiesced snapshots?

edit: we replicate by using XenServer API which is on block level, so we don't mess with the content at all

buzzzo commented 7 years ago

I suppose the snapshot is orchestrated by xenorchestra and thus I don't see any settings on the replication task which indicates if is it quescied or not. On the xencenter console on the snapshot tab the snap taken by the replication job indicates on the tags the value "quiesce" so i would assume the snap is quiescied.

Thx

olivierlambert commented 7 years ago

You can see if a snapshot is quiesce in XO: go into the VM, Snapshot tab. There is a small icon if it's quiesced.

I'll try to see if I can replicate the issue, but as I said, we don't have any clue of what's inside the disk itself, we just export the data to another XenServer, without interfering at all inside.

buzzzo commented 7 years ago

I can confirm: is quiesced. Just to try out if by disabling quescied snap the replicate works: is it possible to specity the snap type in the replication job ? thx

olivierlambert commented 7 years ago

You can't disable quiesce directly, but you can stop the VSS service it will generate a non-quisced snapshot. But I'm not sure it will solve your issue. I don't know if it's related to that, in theory a quiesced snap should be more healthy. Can you check the source VM logs, if you see a VSS call on all drives?

olivierlambert commented 6 years ago

Closing due to inactivity. Feel free to comment if it's still a problem.

buzzzo commented 6 years ago

NOTE: it's solved for me too (switched to alike and replicate correctly)

olivierlambert commented 6 years ago

DId you still use VSS or agent based replication? Because if it's agentless, it's exactly the same way for us to do it.