[
https://issues.apache.org/jira/browse/CLOUDSTACK-8762?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14718259#comment-14718259
]
ASF GitHub Bot commented on CLOUDSTACK-8762:
--------------------------------------------
Github user bhaisaab commented on the pull request:
https://github.com/apache/cloudstack/pull/754#issuecomment-135709691
@wido just checked again, most debug messages are in loop and changing them
to info or errors would be unnecessary. In case of error, run time exceptions
are thrown that would be captured in the logs and won't allow VMs to start.
@miguelaferreira I discussed the issue of changing logic there, it seems
that the mtime should be reliable enough and I won't be adding md5 or other
checksum checking logic as that will be too slow and costing on CPU and IO. Do
you still want me to refactor the logic as a separate class or move to say
FileUtil in the cloud-utils package?
> Check KVM disk files for activity before starting VM
> ----------------------------------------------------
>
> Key: CLOUDSTACK-8762
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8762
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Reporter: Rohit Yadav
> Assignee: Rohit Yadav
> Fix For: 4.5.3, 4.6.0
>
>
> KVM hosts have real issue with fencing when storage are accessible by
> multiple hosts such as in the case of NFS, in which case there is a
> possibility of qcow2 disks being accessed by multiple hosts causing disk
> corruptions. The aim of this ticket is to explore ways to delay starting a VM
> if the qcow2 disks/files are still being accessed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
