[jira] [Commented] (CLOUDSTACK-379) No (simple) way to recover if virtual router gets corrupted in a basic zone

["Prasanna Santhanam (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CLOUDSTACK-379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13479905#comment-13479905 ] 

Prasanna Santhanam commented on CLOUDSTACK-379:
-----------------------------------------------

Worth saving the workaround for pre-4.0.0 deployments:


<clayb> Would it be worthy of a bug to raise that if my virtual router's root disk gets corrupt in a basic zone, I don't believe there's any way to recover? (Since if I destroy the instance, I can't restart the network to get it back -- as I understand it.)
<clayb> I'm not finding an obvious way to work around my virtual router needing to be fsck'd
<clayb> (Darn NFS drop-outs!)
<rohityadav> clayb: if you're using ASF master, the restartNetwork is fixed now… you can goto Network and do a restart with cleanup enabled or delete the vr from infrastructure and restart Network…. if you're using 4.0 or prev. there is no direct way; 1. delete VR, start a new VM, a new VR will be created; or 2. delete the VR from host using xe or xencenter or virsh etc. and hope that after clusterSync HA will recreate a new VR (am not sure about 2, it can take time
<clayb> Okay
<vogxn> rohityadav, how about clayb's suggestion of copying the image from a working VR?
<rohityadav> vogxn: too many troubles, besides VR is stateless, if you destroy it outside CS's scope the HA thread should recreate it, check ha thread interval settings from global settings clayb 
<clayb> If I could get it back by simply starting a VM, that'd be easist
<rohityadav> clayb: yes, kill the old VR, start a new VM and done!
<clayb> Oh, that rocks! I thought destroying it would kill it forever
<vogxn> only i'm wondering if the re-spawn works on new VM start for basic zone. i konw adv. zone it works well
<rohityadav> clayb: yes when admin does destroy VR, it won't be recreated unless a new VM is launched which triggers deploying VR for its basic network; it's a hack for >4.0; we've a better network restart fix for basic zone now
<clayb> Okay, I'm still on 3.0.3
<rohityadav> vogxn: the code is same, so if a new VM is launched findOrDeployVirtualRouter is called for both cases and a VR is deployed in case its missing; applies for both adv and basic zone
<rohityadav> clayb: yes, that hack will work; kill old vr, create new user instance vm
<clayb> Okay cool; trying that
<clayb> Thank you rohityadav and vogxn!
<clayb> All is again happy :)
                
> No (simple) way to recover if virtual router gets corrupted in a basic zone
> ---------------------------------------------------------------------------
>
>                 Key: CLOUDSTACK-379
>                 URL: https://issues.apache.org/jira/browse/CLOUDSTACK-379
>             Project: CloudStack
>          Issue Type: Bug
>          Components: Network Controller
>    Affects Versions: pre-4.0.0
>         Environment: Cloud Stack 3.0.2 on CentOS 6.3
>            Reporter: Clay B.
>            Priority: Minor
>
> If the virtual router gets corrupted in a basic zone, the UI and API options for recovery are very difficult, requiring one to delete the zone and start over.
> For an advanced network, one can simply follow: http://docs.cloudstack.org/Knowledge_Base/Recovering_a_Lost_Virtual_Router
> However, for a basic zone, deleting the

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira