VMware

VM failure to Power On

Hello all,

One of our sysadmins failed to remove a snapshot from before the weekend and we’ve come in today to find that the server is hard down and unable to be brought back online. The specific admin snap’d the server on last Thursday for windows patching, checked “snap server memory” and it looks to have ballooned up to about 750+ GB. Right now, all 4 disks on the VM are mounted with the 000001.vmdk instead of their flat .vmdk files. All 4 of the disks are pointed towards different datastores (compellent SAN i know ugh).

Here is the error from vcenter events:

>Error stack:
>
>Module ‘Disk’ power on failed.
>
>Cannot open the disk ‘/vmfs/volumes/uniqueidentifier/servername/servername-000001.vmdk’ or one of the snapshot disks it depends on.
>
>There is not enough space on the file system for the selected operation
>
>File system specific implementation of FileIO[file] failed
>
>File system specific implementation of Ioctl[file] failed
>
>File system specific implementation of FileIO[file] failed
>
>File system specific implementation of FileIO[file] failed
>
>Failed to close file with handle 1234

He is advising that we cannot delete the snapshot as the windows patching seems to have broken some things on the server and needs it just in case those issues cannot be fixed. What are our best options here to get the server back up? Should we increase the datastore size to accommodate the snapshot so we can boot the VM, or is deleting the snap the best option?


View Reddit by coreywaslegendView Source

Related Articles

One Comment

  1. Figuring out how to free up space will be really important. Extending the datastore, if possible, is one way to go.
    Another thing to consider is actually shutting down another VM on the datastore. Running VMs have vswap files that take up space when they start. By shutting down another VM on the datastore, that space will be freed up. Though, depending on how much storage is needed to start that VM, you may need to shutdown multiple VMs.
    The biggest consideration, though, is why the snaps need to stay. Sounds like whatever space you can provide to the datastore is going to be eaten up pretty quickly. This could be service impacting for other VMs on the datastore as well. Fixing the issue within the VM is really important as you cannot easily reclaim the storage you may allocated to extend the datastore.

Leave a Reply

Your email address will not be published. Required fields are marked *

Close