[
https://issues.apache.org/jira/browse/CLOUDSTACK-8129?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Likitha Shetty resolved CLOUDSTACK-8129.
----------------------------------------
Resolution: Fixed
+Root Cause Analysis+
This issue will only happen in case VMs are migrated across VMware DCs in a
legacy zone because in case VMs are cold-migrated within a DC, CS ensure the
stale VM is cleaned up once the new VM is successfully brought up. But since we
can look-up a VM only within the scope of VMware DC, CS cannot clean up the
stale VM from a different DC.
+Solution+
If VM has been cold migrated across different VMware DCs, once the VM cold
migration is complete, unregister the VM from source host and cleanup all its
artifacts.
Also, through UI we cannot migrate a volume attached to a stopped VM across
clusters. But with the API we can. In case of legacy zones, users could
potentially use the API to migrate a ROOT volume attached to a stopped VM
across VMware DCs. Then we will run into the same issue reported in this bug.
Hence added an additional check to block the following use-case: Volume
migration of ROOT volumes attached to a stopped VM across VMware Datacenters
for legacy zones. Users can instead achieve this use case by migrating the VM
to the storage pool itself.
> [VMware] Cold migration of VM across VMware DCs leaves the VM behind in the
> source host.
> ----------------------------------------------------------------------------------------
>
> Key: CLOUDSTACK-8129
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8129
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Affects Versions: 4.5.0
> Reporter: Likitha Shetty
> Assignee: Likitha Shetty
> Priority: Critical
> Fix For: Future
>
>
> 1. Have an upgraded VMware setup with a legacy zone (Legacy zone is a zone
> where a CS zone has clusters that are spread across different VMware DCs). So
> ensure there are two clusters say C1 and C2 that belong to VMware DCs D1 and
> D2 respectively.
> 2. Deploy a VM. Say it is in C1.
> 3. Stop the VM.
> 3. Migrate the VM to a storage in a C2.
> 4. Start the VM.
> Observations - VM will be be successfully started in a host in C2. But a
> stale entry of the VM will be left in C1. And whenever VM sync kicks in it
> stops ans start the VM repeatedly because there are two VMs by the same name
> in vCenter.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
