Datastore Filling Automatically

[not_today88]

Hi All, 

I'm fairly new to VMware and have searched for similar questions to no avail. My issue is when I create a new thick provisioned datastore and then go into the VM settings to increase the disk space, the datastore automatically expands to nearly 100% capacity, rendering it unusable until I provision more space. For some reason the .vmdk file appears to be doing this, but I don't understand why. This is a brand new VM with no data stored on it yet. There are no snapshots either. The server was just built yesterday. 

A) How can I correct this so I don't waste space?

B) How can I prevent this from happening in the first place? When comparing other multiple drive VM's, the drives are also thick provisioned, lazy zeroed but they don't seem to have this issue. 

Stupid question, but in Windows disk management, it still shows plenty of free space because it's not been used. Does this mean the server will still function okay as long as there is free space available on the datastore? Apparently what is shown as used in vSphere 6.5 and Windows is not the same or related. This is a new SQL server so it's critical that it be reliable. Thanks in advance. 

[scott28tt]

Can you be clear on the terms in your description of the issue, I'd suggest using these terms:

Datastore = the unit of storage used by ESXi - a VMFS partition or NFS volume on whatever underlying physical storage you use

Virtual Disk = the VMDK file for a VM - this is the thing that you define in vSphere as thick or thin provisioned if using VMFS

Guest Disk = the view that Windows has of the Virtual Disk (as seen in Disk Management)

Guest Partition = the Windows partition inside the Guest Disk (as seen in Disk Management)

[not_today88]

Thanks, Scott. My apologies for not using the correct terminology. I believe my concerns have been addressed but for future reference, why does a thick provisioned virtual disk expand to the full capacity of the datastore when I increase the size in the disk settings? I understand thick provisioning does this by design, but when it does so, it created a critical alert for the datastore and in come cases, I'm unable to boot the VM. Is there something I'm doing wrong and can do to prevent this? 

Thank you for your time. I'm still learning VMware (obviously). 

[pdirmann01]

How big are you making the new VMDK, how big is your datastore, and how much is already consumed (keyword being consumed, not allocated)? Remember that thick provisioning will consume the space immediately. If you have 1TB datastore with 500GB already consumed, deploying a thick provisioned 500GB drive will fill your datastore up.

[not_today88]

I just built a SQL server with multiple disks and a dedicated datastore for each one. For the guest data disk, I started with a 700GB datastore in vSphere (6.5). After I attached the disk to the VM I increased the space to 700GB. And then the datastore subsequently filled up to 700GB, setting off a critical alert. I then had to provision another 100GB from the SAN to make the datastore happy. I was then able to adjust the guest disk in Windows and all is good, but it seems I shouldn't have to do that and I'm missing a step. 

[scott28tt]

So a 700GB datastore containing a 700GB thick-provisioned virtual disk.

Yes, that would absolutely fill the datastore.

What were you expecting to be different?

Why use a dedicated datastore for every virtual disk?

[not_today88]

Sorry, I forgot to mention I originally only increased the disk by half, but the datastore increased 100%. Once I added more space to the datastore, the alert cleared and I increased it to 700. 

In any case, my server is fine for now and I'll review the official documentation. I was going off of what we have in-house and it may not be best practice.