Description of problem: ----------------------- For RHEV-RHGS HCI disaster recovery, strategy recommended was to geo-replicate the data domain after snapshotting the VM disks, to the central site (i.e) geo-replicate ( MASTER ) at Remote site to another gluster volume (SLAVE ) at the central site. After importing the geo-replicated data domain at the central site, unable to list the unregistered disks. The disks needs registering, in order to list the disks as the floating disks, and then it could be attached to the App VMs Version-Release number of selected component (if applicable): -------------------------------------------------------------- RHGS 3.1.3 nightly ( glusterfs-3.7.9-3.el7rhgs ) RHEV 3.6.5 RHEL 7.2 How reproducible: ----------------- Always Steps to Reproduce: ------------------- 1. Create a data domain backed by gluster sharded replica 3 volume 2. Create VMs with their additional disks part of the above created data domain 3. Take snapshot of the VM's disk 4. Checkpoint the geo-rep session and start the geo-replication of data domain 5. After geo-replication reached checkpoint, stop geo-replication [Note - step4 & step5 done via geo-rep scheduler script] 6. Import the slave data domain in another RHEVM instance 7. Try to list the unregistered disks Actual results: --------------- snapshotted disks are not listed Expected results: ----------------- snapshotted disks should be listed so as the help in registering the disks after fetching the disk id Additional info: ---------------- I have deleted the the overlay image file from the data domain, leaving only the backing image. Now I could list the unregistered disks
Any logs? Tal - do we know what the issue is?
(In reply to Yaniv Kaul from comment #1) > Tal - do we know what the issue is? No logs - no analysis ;-)
The issue is to do with listing unregistered disks in case disk has snapshots. Currently such disks are skipped. As per Maor, this is an RFE to manage floating disk with snapshots - changing title to reflect same.
*** Bug 1056944 has been marked as a duplicate of this bug. ***
This was implemented in a different way for the HCI use case. If relevant, please re-open and explain why. As always- patches are welcomed!