The operation on file "/vmfs/devices/deltadisks/test.vmdk" failed. ESXi 6.0

[gmoro]

Hi,

I'm using a image created using qemu-img and I get "The operation on file "/vmfs/devices/deltadisks/test.vmdk" failed. ...." in the WebUI and I can see this errors in VMWare log in the host

2018-01-18T14:17:51.553Z| Worker#0| I120: FileIOErrno2Result: Unexpected errno=12, Cannot allocate memory 

2018-01-18T14:17:51.553Z| vmx| I120: VMXAIOMGR: Retry on read "/vmfs/devices/deltadisks/2828f296-test-s001.vmdk" : Cannot allocate memory. 

2018-01-18T14:17:51.553Z| vmx| I120: VMXAIOMGR: system : err=c0002 errCode=12 freeSpace=0 

2018-01-18T14:17:51.553Z| vmx| I120: VMXAIOMGR: "/vmfs/devices/deltadisks/2828f296-test-s001.vmdk" : read s=42949279744 n=131072 ne=1, fai=0 

2018-01-18T14:17:51.553Z| vmx| I120: VMXAIOMGR: v[0]=B802B9B8000:131072 

The image sometimes start the boot process and hang before is fully booted, other times it gives this error immediately before even GRUB finished loading.

Using vmkfstools -x check show no errors in the disk image. Also importing the image makes the image work, but I want to get the image working from the original.

For context, the image is using the IDE controller, and was created with monolithicSparse type. One other detail is that when I run the image the system split the disk  metadata automatically moving the main image to a file suffixed with "-s001" in the file name and it changes the metadata from createType="monolithicSparse" to createType="twoGbMaxExtentSparse".

Does anyone have any pointers on how to figure out what the problem is exactly?

[msripada]

Please check the KB article as the error out of memory messages is similar to the below kb

FileIOErrno2Result: Unexpected errno=12, Cannot allocate memory

VMware Knowledge Base

Thanks,

MS

[gmoro]

Thanks for the answer, the system is a test system has no other VM's running at the moment, and I checked the heap size, is in the default of 256 mb

[msripada]

Can you check the vmkernel and hostd log when trying to power on ..which can give us some more information about why it is failing to power on

Thanks,

MS

[continuum]

Hi
I have seen this quite often last months.
The trick is to clone the disk once via vmkfstools -i current.vmdk new.vmdk
ESXi can handle this type of vmdks but will not create snapshots - so backing up such a disk via Veeam&Co is impossible.
Once converted into standard "VMFS" format and your problems will be history.
Ulli

[gmoro]

Attached the relevant parts of the log

[gmoro]

Hi

I wanted to avoid the cloning, qemu-img should be able to create a compatible images, otherwise a fix should be made on it, in my use case I can't really depend on the vkmfstools

[continuum]

quemu-img creates a descriptorfile that will not work. It will create a flat.vmdk that is fine but the descriptor needs manual editing.