VMware Horizon Community
btrabue
Enthusiast
Enthusiast

Insufficient Disk Space on DataStore error when recomposing

I have a very urgent need in some help.  I scheduled a recompose to happen this weekend and I did not notice that my replica disk that my master image was sitting on is down to about 28GB.  When the recompose started on the first VM I received the insufficient disk space error.  I have about 75 VM's in this LC pool and they are all on various snapshots.  All of the snapshots that these VM's are on no longer exist with the master image.  I have been told to never have more than 2 snapshots per image due to storage space on the SAN.  I need to recompose them so that I can then perform a rebalance.

What are my options here?  I really need to try and recompose these VM's with the new snapshot this weekend.  I have looked at the replica disk and I am not sure what can be deleted to get more immediate space on the disk.  Are there any other options that I am not aware of?

Thank you in advance for your help!!

Reply
0 Kudos
8 Replies
psplnalin123
Enthusiast
Enthusiast

Just try to delete the files where your VDIs reside right now there is no sufficient space so delte the files. Are you using persistent vdis or nont persistent vdis

Reply
0 Kudos
btrabue
Enthusiast
Enthusiast

What files do I need to delete? When I browse the datastore there is nothing but replica folders with weird names that I am not sure what they are used for. We are using non persistent vdi's.

Reply
0 Kudos
psplnalin123
Enthusiast
Enthusiast

ngare you using same datastore for Replica and Non persistent VDIs

Non Persisten VDIs are in refresh mode ?

I Hope non persistent VDIs are not in refresh mode . If not then change the pool settings change it to refresh mode and login to vdis one by one on 5-7 vdis and then recompose.

Reply
0 Kudos
btrabue
Enthusiast
Enthusiast

I tried to refresh a VM yesterday and it is in an error status now because the snapshot that the VM was using cannot be located. I need to perform a recompose to get all of the VM's at the same snapshot.

Reply
0 Kudos
psplnalin123
Enthusiast
Enthusiast

Okey seems like that you have deleted earlier snapshots. To resolve this you need to change the pool setting to refresh mode. Secondly you need to delete the files where your Non persistent VDIs reside.

Total how many VDIs in that pool can we delelte 3-4 VDIs and immedite we could do the recompose

Reply
0 Kudos
btrabue
Enthusiast
Enthusiast

So, can I refresh just a couple of VM’s at a time instead of doing the whole pool? Every VM in the pool is assigned to a user. I really don’t have any VM’s that can be deleted.

Reply
0 Kudos
gunnarb
Expert
Expert

I do want to clarify that I think you were misled about snapshots.

On you master it doesn't really matter how many snapshots you have (so long as you don't go beyond the maximum).  The reason I say this is that your Master is never used.  Its not like your Server Infrastructure, where you do want to be very careful with snapshots.  With SI, those VMs are actively used so you shouldn't ever run on snapshots.  With VDI, since the master isn't ever used, except when you update it for the next recompose, you can't actually have quite a few snapshots.  Now its still good practice to go back and clean up those snaps once you have no need of them.  But I recommend my clients utilize snapshots to have a history of the master image.  What you need to be careful about is replicas.  These are clones of your master's snapshots, and in a clean environment you should only have one replice per datastore (if you only have one master image per datastore).  The problem many of my clients have is that they will end up recomposing but can't recompose a few VMs so they end up stuck on an old snapshot, which leaves that replica out there.  Replicas are the thing you really need to watch, snaps on your master aren't that big of a deal.

And as you have seend, when you delete your snaps you have to go into your pools and make sure the paths are updated, otherwise when you recompose it will start failing.  This really hurts when administrators make this mistake in environments that are set to recompose after a logout.

Gunnar

Gunnar Berger http://www.gunnarberger.com http://www.endusercomputing.com
btrabue
Enthusiast
Enthusiast

After being on the phone with VMWare we got it figured out.  I ended up having to add about 100GB to the replica disk so that I could recompose the pool.  After recomposing the pool and getting everyone onto the same image I cleared up over 150GB of free space on the disk.  Lesson learned, try to keep all of your users in a pool on the same image.  I was updating the image from time to time and creating new VM's with the updated image when a new VM was requested.   If I did not have the option to add 100GB to the drive I was going to have to create a new pool and move everyone's UDD to the new pool and that would have taken a lot more work.  Thanks for everyone's responses!

Reply
0 Kudos