Unsolved
This post is more than 5 years old
4 Posts
0
2288
July 28th, 2019 19:00
Utilization at ScaleIO level is higher the total utilization at OS level
We have configured thin luns in scaleIO and our scaleIO usage is above 80% .We ran "esxcli unmap" command to recover space and recovered few TBs but still the utilization is higher than we have expected . We ran OS commands to get file system utilization on all severs but total utilization via this is very less as compared usage shown at scaleio level .
Is it because thin provisioing is not working as expected as ideally scaleio usage should be more or less same as total OS level utilization as we are doing thin provisioing ?
Please help or suggest what shall we do to recover the space .As of now we are gettting thin provisioing savings and the luns are behaving as .
Our ESXi version is 6.0
Scaleio version is 2.0.-13000.211
matt_hobbs
31 Posts
0
July 30th, 2019 05:00
ScaleIO thin-provisioning has always worked well in my opinion --- it might be worthwhile for you to run a LiveOptics assessment over this environment if you can. https://www.liveoptics.com/
Also please see this article as it might relate to your particular situation: https://support.liveoptics.com/hc/en-us/articles/360009480413-VMware-VM-Occupancy
storage_support
4 Posts
0
August 5th, 2019 01:00
Thanks a lot for response .The link related to "VM occupancy" is not the solution .My scleio luns are thick but vmdks are provisioned "Thick provision lazy zeroed" and ideally as an example if lun size is 100 and datastore created on it is 100 GB and assigned 60 gb vmdk to OS with OS utilization a 20 GB then scaleio usage should be 20 GB but in my case it is showing much greater than 20 GB,
Can you please guide how i can use live optics for assessment .It have various options and i am not sure which one to use
matt_hobbs
31 Posts
0
August 5th, 2019 10:00
matt_hobbs
31 Posts
0
August 5th, 2019 10:00