Snapshot error

[pizang]

After upgrade to vSphere and ESXi4 I have problem with making snaphots for 2 VMs.

One machine has 2 disk

system disk of size 10 GB - 175 GB of free space

data disk of 500GB - 200MB of free space

When I try to make a snapshot an erorr appears:

File <unspecified filename> is larger than the maximum size

supported by datastore '<unspecified datastore>

I thought it was a problem with block size (2MB) that is limiting file size to 0,5TB. I have created a new LUN the same size as the old one - 512GB. I have

created a new VMFS volume with 4MB block size (so it can hold files as

big as 1TB). I have powered down my VM and moved disk 2 (data disk) to

new VMFS.

To my surprise I received exactly the same error.

Is, by any means, block size of configuration file's partition involved?

What I mean is that VM configuration file and VM system disk is stored

on partition with block size of 1MB. I have checked and only two VM has

problem with snapshot and both of them (and only them) have virtual

disks larger than 256GB. As snapshot file is stored on disk where

configuration file is located maybe it can cause a problem? Maybe

system is checking block size of snapshot partition?

[pizang]

I have resolved this. I have moved VM config file to VMFS with 8MB block. Snapshot was created without any error. So ESXi 4 changed the rules. It is checking whether snapshot destinantion VMFS has block size equal or grater than disk that you are snapshotting.

[depping]

Yes this seems to be a new check indeed. It makes sense though cause the snapshot could potentially grow beyond the max file size limit dictated by your VMFS block size.

Duncan

VMware Communities User Moderator | VCP | VCDX

-

---

Blogging:

Twitter: (*NEW*)

New book in town: vSphere Quick Start Guide ()

If you find this information useful, please award points for "correct" or "helpful".

[pizang]

This means that your snaphot should be at least bigger than 256GB - whoa! I would never let snaphot to grow so much

It would be nice to have this change stated somewhere in upgrade documentation.

Thanks for confirmation!

[depping]

I know normally no one would do this, but it's possible... and we want to prevent this from happening and causing issues because of the block size.

Duncan

VMware Communities User Moderator | VCP | VCDX

-

---

Blogging:

Twitter: (*NEW*)

New book in town: vSphere Quick Start Guide ()

If you find this information useful, please award points for "correct" or "helpful".

[ITQPG]

I don't get it.

Should it not check for VMFS size (as well).

You can have the right blocksize but not enough free space left.

Or could it be that this is less of an impact because, as long as the blocksize is the same or bigger, the last block written is always a whole block. which doesnot cause you to lose data...?.

Mmmmm, did i just get it ?

[xlpnycm]

Hi,

I have ran into the same issue and confirmed that the problem is the datastore blocksize.

I have recently upgraded from ESX 3.5 to ESX 4. This problem didn't occur in ESX 3.5. Do You know if there are any plans to modify this behavior back as it was in ESX 3.5?? (a PTF, configurable parameter, etc ...)

Your comments will be greatly appreciated

Thanks in advance.

Xavier