[
https://issues.apache.org/jira/browse/CLOUDSTACK-8985?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14971621#comment-14971621
]
ASF GitHub Bot commented on CLOUDSTACK-8985:
--------------------------------------------
Github user mike-tutkowski commented on the pull request:
https://github.com/apache/cloudstack/pull/968#issuecomment-150663033
@DaanHoogland At present, I don't have any way to run SolidFire integration
tests inside of the Apache community. Perhaps I can leverage a virtual
appliance at some point to make this happen, though.
@pdube It turns out it's actually the way the storage framework is
implemented. When it invokes the "delete" method on the storage plug-in, the
"removed" column has not yet been updated in the database. That only happens
once the plug-in returns "success" on its side (then a callback method is
invoked within the storage framework to update the "removed" column (among
other things)). My assumption in the plug-in was initially that this "removed"
column was updated before the storage framework told me to remove the object,
but that was not correct (so I need to treat the volume like it's still "alive
and well" and just ignore it when I calculate space numbers).
> Deleted volume's removed column not updated
> -------------------------------------------
>
> Key: CLOUDSTACK-8985
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8985
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: Management Server
> Affects Versions: 4.6.0
> Environment: All
> Reporter: Mike Tutkowski
> Assignee: Mike Tutkowski
> Fix For: 4.6.0
>
>
> Note: Relates to managed storage only.
> Due to the fact that a volume does not enter the "Expunged" state until after
> a callback method is invoked (which happens after the plug-in is notified to
> delete the volume in question), the current getUsedBytes implementation (in
> SolidFirePrimaryDataStoreDriver) was assuming the volume just deleted would
> not be included in the used-byte count calculation.
> The fact that the code does try to count this volume in the used-byte count
> calculation leads to an exception being thrown and the removed column not
> being updated.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
