cloudstack-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Likitha Shetty (JIRA)" <j...@apache.org>
Subject [jira] [Resolved] (CLOUDSTACK-8125) [VMware] After cold migration of a VM, VM starts with warning messages in MS logs
Date Wed, 24 Dec 2014 09:21:13 GMT

     [ https://issues.apache.org/jira/browse/CLOUDSTACK-8125?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Likitha Shetty resolved CLOUDSTACK-8125.
----------------------------------------
    Resolution: Fixed

+Root Cause Analysis+
Things to note
* After VM migration, ROOT volumes's VMDK name changes. And in the _volumes_ table in DB,
_path_ is updated to reflect this change but _chain___info_ is not.
* In case of VMware, when VM is cold migrated, vCenter is unaware of the migration and hence
unaware of the change in the location of the VM's volumes.
* In case of VMware, when CS attempts to start a VM, we retrieve VM's current disk information
from vCenter, tear down the disks from the VM, re-configure the VM with the disks and power
on the VM. During VM re-configuration with the disks if disk information is returned by vCenter
in the first step, then that information is honored and used to re-configure the VM. Otherwise
CS build the disk information based on the volume's path.

What happened during VM start after a cold migration?
VM was cold migrated and vCenter was unaware of the change in the ROOT volumes's location/path.
And so when asked for VM's disk information, vCenter returned the old path i.e. the path before
the VM was migrated. CS then re-configured the VM with this wrong disk information and tried
to power-on the VM. This power-on failed with the warning message.

But CS always retries if VM fails to start on the first attempt. And during the second attempt,
when asked for the disk information, vCenter returned with no disks. vCenter returned no disks
because CS tore down the disks during the first attempt to start the VM. Because vCenter didn't
return any disk information, CS built the disk information based on the volume path (which
had been rightly updated after the cold migration), re-configured the VM with the right disk
and successfully powered-on the VM.

> [VMware] After cold migration of a VM, VM starts with warning messages in MS logs
> ---------------------------------------------------------------------------------
>
>                 Key: CLOUDSTACK-8125
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8125
>             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
>
>
> +Steps to reproduce+
> 1. Deploy a vm.
> 2. Stop the VM.
> 3. Add another primary storage to the same cluster.
> 4. Migrate the VM to the new primary storage.
> 5. Try to start the VM.
> Issue - First attempt to start the VM fails with
> {noformat} 
> java.lang.RuntimeException: File []/vmfs/volumes/27951060-bbb89fa5/i-2-10-VM/daadfb44467f41e6a38fd8ba05e28cb4.vmdk
was not found
> {noformat}
> The VM eventually starts up but the disks are migrated back to the old storage.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message