Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

what I can figure out is by default when we spin up VM's in openstack the disk is always thin-provisioned i.e. if I create a VM having a disk of 200GB then it will not create a QEMU/KVM disk of 200GB for the instance, but rather it will increase the size of the file which is the disk of the VM as and when it starts getting occupied.

Even I am running the latest Mitaka release and was able to create 3 VM's having hard disk of 500GB each but my compute node had only 120GB of physical hard disk :P .

Now the question is weather to treat this as a bug or not as issue will not be at this point of time but say in my case when the VM's actually start occupying the space then they would need a total of 1.5TB of space (500GB for each VM)...which would ultimately fail as the actual space is only 120GB ...

One solution is to always create VM's having a thick provisioned disk where the amount of Hard disk size will be preallocated weather is again not a best idea

what I can figure out is by default when we spin up VM's in openstack the disk is always thin-provisioned i.e. if I create a VM having a disk of 200GB then it will not create a QEMU/KVM disk of 200GB for the instance, but rather it will increase the size of the file which is the disk of the VM as and when it starts getting occupied.

Even I am running the latest Mitaka release and was able to create 3 VM's having hard disk of 500GB each but my compute node had only 120GB of physical hard disk :P .

Now the question is weather to treat this as a bug or not as issue will not be at this point of time but say in my case when the VM's actually start occupying the space then they would need a total of 1.5TB of space (500GB for each VM)...which would ultimately fail as the actual space is only 120GB ...

See in my case as well image description

One solution is to always create VM's having a thick provisioned disk where the amount of Hard disk size will be preallocated weather is again not a best idea