Datastore Cluster and a big disk

Hi guys,

i have a datastore cluster with a whole heap of 4TB Luns, each has around 1.5TB free so fairly comfortable space wise.

i need to make a new 2TB disk for a server, but it seems i'm at a loss as to how to do this..  there is 30TB free in the cluster itself so heaps of space but vsphere wont let me add the disk as each individual LUN doesnt have the required space..

Any tips on how to get around this please...  (last time i moved a datastore out of the cluster put the disk on it then moved it back in.. surely that cant be the best way!)

0 Kudos
3 Replies

Hi there AllBlackCats,

Couple of options that I can think of:

  1. Move VMDKs/VMs around so that you have a data store with more than 2TB available
  2. Put a couple of data stores into maintenance mode and use these to create a single larger data store

You do have another option but it is a bad bad idea - create two VMDKs and join these within the guest OS. Only putting it here for completeness but would NEVER recommend anybody do this in production.

Kind regards.

0 Kudos

> You do have another option but it is a bad bad idea - create two VMDKs and join these within the guest OS.
I completely agree that this is a bad idea - however there is one that is worth mentioning ....
Lets say you need a disk with driveletter E: and a total of 4.5 TB
The E: drive has 3 subdirs like
To acchieve this it would be safe to create 3 VMDKs with a size of 1.5 TB each.
The first one is mounted by Windows to driveletter E:
The second one is mounted by Windows to the directory E:\development
The third one is then finally mounted to the directory E:\archive.
This is still an ugly solution but it is quite safe.
I would only recommend doing something like that if you document your VMs REALLY accurately.
If not - the next admin that has to handle this VM will hate you.
And to add to what Glen already said: dont use Windows dynamic disks to create large guest drives.

Do you need support with a recovery problem ? - call me via skype "sanbarrow"
0 Kudos

That’s funny - we use mount points at work and I completely forgot they were an option as well. The main reason we did it is legacy, I.e. before greater than 2TB VMDK.

While we could revisit and create a single drive of multi TB in size there are still advantages to leaving it this way. The biggest is backup as we can treat each folder (Mount Point) as an individual object.

continuum is right however that it needs to be documented. Even then I have lost count of the number of times people get confused with what is happening. For Linux people this is probably considered normal and for those of us old enough to remembed ESX Classic this was part of your day. Not anymore and the bemused look on people’s faces is often priceless especially when they check the free space and drive size on the volume these live in, I.e. how can we fit 3TB of data in a 1GB drive Smiley Happy

Also if you are using in-guest agents for backups, make sure if supports mount points. Some backups cannot follow the junction point so either back it up as a file or something else. Monitoring software also needs to understand mount points as it needs to monitor the freespace in the folders not the root system.

0 Kudos