We have somehow ended up with quite a few VMs where storage DRS has been disabled, I presume by people trying to be clever and selecting individual datastores for disks which disables it.
We have some VMs with many disks, more than can fit into a single datastore so they are spread out across the sdrs cluster. What is the behaviour if I go and reenable sdrs on the vm, with the default option of keep the disks together? Will it straight away try and move things around and keep the disks in the same datastore, or does it only come into play on initial placement?
If, due to lack of space in a datastore, will it block addition of new disks to a vm if it can't keep it in the same datastore, or is it only a best efforts basis where it will only keep them together if possible but put them in another disk in the cluster if not?
VMDK keeps together: This is default rule. It keeps all the virtual machine VMDKs on the same datastore (i.e. Under virtual machine working directory). It does not move the vmdk's between data clusters, It is between the same SDRS cluster. For more information, see KB 2149938 and KB 2053792. Also, see https://communities.vmware.com/thread/460195.
Sorry I meant how does it work within the same datastore cluster.
I have a VM with 10 disks, totalling 10 TBs, split across 5 2TB datastores all within the same datastore cluster. This VM has a VM Override set up in Storage DRS to disable storage DRS.
If I then delete the VM override, does it try and shift the disks into the same datastore? Even though it could fill a datastore up? Also, would it complain and throw an error if I need to add a disk later on, and need to put it into another different datastore? How does storage DRS and 'keep vmdks together' rule behave with vms with many disks requiring splitting across many datastores in the storage cluster?
sDRS will validate the move before trying to perform it, so you wouldnt get all 10 disks trying to fit on one of your 2TB LUNs, taking it offline
If the vms virtual disk violates the rule, sDRS makes migration recommendations to correct the error or reports the violation as a fault if it cannot make a recommendation that will correct the error.