Snapshot Consolidation Issues - No Space Available

[jssanche]

Hello!

Your help please. This is a simple scenario, I have an ESXi 6.0 (no vCenter, No shared storage, free version) running a single virtual machine (WINDOWS 2012). This ESXi has a 1 TB datastore, that is almost occupied (96%) by the virtual machine and 3 snapshots. When trying to run a snapshot consolidation, the process fails (no disk space available), so I add a new datastore of 1 TB, to try again to run the consolidation again. My question is: Could the consolidation process detect this another datastore and use the space available there in order to complete the consolidation process OK?, or vSphere will not be able to use the space in the second datastore.

Now I think a better idea is to create an extension of the first datastore, bringing it to 2TB, just before running the consolidation process again.

Thanks in advance

[HassanAlKak88]

Hello,

First, If you tried this all during the vm online, power it off and try again. Online delete of snapshots needs free space on datastore.

Otherwise, sure you have to extend the existing/first datastore.

[a_p_]

Please don't try to run the consolidation again at this point, as this could make things even worse!

Instead enable SSH on the ESXi host, and use putty to connect to it. Then change to the VM's folder/directory, and run

ls -elisa > filelist txt

then download the newly created filelist,.txt along with the VM's configuration (.vmx) file to your PC, compress/zip these two files, and attach the .zip archive to a reply post.

Are there other virtual machines stores on this datastore?

André

[sean0brien]

An extent may be the best option in you use case, although you then have a single point of failure for both datastores(if no disk mirroring, which it doesn’t sound like). 

Another option, if you can afford downtime would be to power down a VM, unregister it from the inventory, relocate VM files to other datastore , register VM and power back up.  At this point you should be able to consolidate.

[continuum]

> My question is: Could the consolidation process detect this another datastore and use the space available there in order to complete the consolidation process OK?

No - thats why Andre asked for the filelist.

Once we got the list we can suggest a procedure that will consolidate the snapshots into a new file on the new datastore.

Please forget about adding the other datastore as an extent to expand the original datastore.

That would be possible but usually we think the risks are too high to justify such suicide-workarounds.

[jssanche]

Hello!

This is how my vm's objects looks like...

Thanks for your prompt reply.

[jssanche]

Hello Continuum!

Why to forget extending my main datastore to 2 TB?

There is a single virtual machine, and all the available space is for this VM, the database inside continues growing day by day. I want all this space for this VM (2 TB), but I wonder why it's too risky to extend the main datastore in my particular case.

Thank you

[jssanche]

Hello a.p.!

There's not other objects in this datastore, this datastore is for this VM and all available space too.

Thanks a lot for your prompt reply.

[continuum]

If you use an extent (thats another available disk) to expand an existing datastore you will get the additional free space - but now you have a construction that is quite fragile.

If one of the physical disks fails the complete datastore is unreadable for 99.9 % of the VMware admins.

We suggest that you add the disk and create a second datastore instead.

Then you can use one of them for the first 2 vmdks if your VM and the other one for the rest of the vmdks.

[continuum]

Next steps :

create the new, second datastore.

Create a new directory inside the new datastore.

Then we have 2 options:

1. move just one larger file to the new directory - so that we have enough free space on the original datastore to consolidate while the VM runs.

This is a bit more advanced than option 2

2.
- consolidate 2 of the  vmdks to the new directory via vmkfstools -i (while the VM is powered off.
- clean up original datastore

- consolidate the remaining vmdks inside the existing old directory via vmkfstools -i

If you want that we create the commands for you - please install WinSCP and connect to the directory where the VM lives now.

Download all the small vmdks without flat or delta in the name.

Download the vmx-file

Attach the vmx and vmdk descriptorfiles to your next reply.

Do not try to do this downloads with the Datastorebrowser - that will not work.

[jssanche]

Hi continuum!

Attached you'll find samples files needed to create the script or set of commands to consolidate the vmdk and delta vmdk through vmkfstools -i command.

Notes: We're using a sample vm running xubuntu for this, cause at this moment we're unable to get the requested files from the virtual machine. Can this files be useful? 

Thanks a lot for your help and support

[jssanche]

I almost forget!!!, the .vmx requested file.

[continuum]

The files you send are the correct ones - but that case is quite different to the one from your screenshot.

It does not help when I write a command where none of the filenames match ...

Can you maybe attach a vmware.log ?