Guys a newbie in Virtualization needs your help.
I tried converting a physical server with 16GB (Basic/Raid-1) and 370GB (Dynamic/Raid-5) partitions. I opted to increase the size of the 16GB partition to 40GB and the 370GB decreased to 240GB as it only ever uses up to 160GB and also due to the fact that the host was formatted with 1MB blocksize thereby limiting me to 255GB max. I also chose the default thick provisioning. During the conversion, the first partition completed okay but immediately failed on the 2nd one. The conversion status said 'Cannot format volume' while the log highlight offered 'Unable to clone volume D'. Now another issue is that the datastore on my ESX has shrunk by 280GB which is the total of the 2 resized partitions. What is the best way to recover this disk space? I am also thinking the reason why volume D failed is because of the data that could possibly be residing after the 1st 240GB on the volume (pardon me if this a farfetched idea). I am trying to defragment the volume anyway. But before I retry the conversion, I want to make sure that I wont run to the same issue again and also have the disk space reclaimed first. So I hope you can give some valuable inputs to do this right. Thanking you in advance.
Hello, I Think that you need more space, you can't use entire datastore, at least 10% should be free. As you know, the swap file reside in Datastore, and need it the same space as VM has configured for memory.
Hi vJJ, appreciate your input.
Let me add some details about my datastore. Prior to the conversion I've got a little over 1TB free out my datastore's total capacity of 1.22TB. Even though the conversion failed, it took up 280GB which is the total space I thick provisioned for volumes C & D. Eventually the free space on my datastore is now 720GB. Are you able to tell me how can I free up that space again? As soon as I sort that one out, I will retry the conversion this time using thin provisioning for the failed volume. The defragmentation went well by the way. Thanks
Just an update: I was able to delete the VMDK file of the failed conversion by going to the Datastore Browser. From the vSphere client, I clicked Configuration -> right-click the lone datastore -> Browse datastore and from there delete the corresponding vmdk file. The free space is back to a little over 1TB now. This is such a simple task likely known by most.
My concern now is whether to choose thin or thick provisioning. Does thin provisioning cause fragmentation of the vmdk file? I can easily choose thick as I expect no more than 4 VMs on the host, each one using around 250GB out of the total 1.2TB datastore on a HP DL380 G7 server. But I would love to hear your expert opinions first.