Good day to you.
We have several compute cluster with their own datastore cluster. We have 10-12 ESXi hosts per cluster and five datastores per datastore cluster.
The storage, a VMax 40k, includes 100TB physical space and our over provisioning is 100%, so wie have 200TB provisioned in datastores (all 2TB sized) in total.
The VCenter is latest 5.5 version and the ESXi is latest 5.1 version.
Ok, now, I ran some reports on the virtualization environment and see, that 50% of the physical space is already used by virtual machines. The report
recognized almost all virtual machines in the environment. But in the opposite, the storage system complains, that we have 15% free, so there are round about 35TB of storage, which is already eaten up and I do not find out, why.
So, what I did:
1. opened ssh one one host per cluster and ran my storage reclaim script
2. The script is:
#!/bin/sh
for datastore in `ls -d /vmfs/volumes/NUE*`
do
echo "Datastore: $datastore"
cd $datastore
echo "Unmap 99% of unused space"
vmkfstools -y 99
sleep 60
done
3. Start reclaim job on the VMax for all thin volumens
4. Wondering, that only up to TB of storage are reclaimed.
Now I am looking for investigation method to examine that problem. Or, am I wrong in my procedure?
Best regards,
Jan
P.S.: I know many links and forum entries here explaining the reclaim procedure, but every time I ran into the problem of missing space, which is not reclaimed.
