vSphere inflating VMDKs/NFS migrations even if explicitly told not to

This post was originally published on this site

I don’t think VMware is going to solve renaming VMs on non-vSAN datastores by now so to rename a new-and-never-turned-on zeroMB VM I migrated it to a temporary datastore to move it back with the new name but it still took about 8 minutes, way too long to move not 1MB of data.


This has been hapenning but I thought it was just me being careless because I was migrating VMs in batches with sleep depravation but this this time I took all precautions and ignored the datastore policies (which would end up in the same result anyway) and set the thin-provisioned disks to migrate thin provisioned. Yet it took a while, it showed it was occupying even less space in the next datastore by fractions of kilobyte, but still, and in reality it inflated a 32GB disk. The machine doesn’t even has an OS in it or even an ISO attached to it.


In the documentation says remote datastores default to thin provisioning, I set thin provisioning on disk, I set thin provisioning on migrations and policies but they keep getting ignored, because most disks in the arrays are flash storage I have made it a point to avoid to choose any option anywhere related to thick provisioning disks, eager or lazy-zeroed. How do I prevent this from happening?


Screen Shot 2020-10-30 at 7.17.30 AM.png

This is on the vSphere 7 btw, the latest, since it now announces updates on right on vCenter the Web Client, you basically cannot be outdated or it’ll keep telling you so.


Message was edited by: Gustavo Domínguez

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.