Unsolved
4 Posts
0
1615
Dell Equal logic PS 5000 Volume space auto growing
All,
I have EQL PS 5000E storage connected to Citrix XEN environment .Since last couple of days, I am observing Volume space is auto growing .
My Citrix Volume size is 7 TB and on Citrix XEN center utlization is showing 4.5TB but on storage GUI it is showing 6.TB .
Looks like Storage is not releasing space. Looking for solution to fix this issue.
dwilliam62
1 Rookie
1 Rookie
•
1.5K Posts
0
May 29th, 2018 07:00
Hello,
That's because Citrix doesn't use a single cluster filesystem that's capable of sending the SCSI UNMAP commands to tell storage what blocks to release back to free space. Without those commands block storage has no way of knowing that a block is no longer needed.
The VMs don't own the disk so they can't send the commands either.
The OS's will reuse previously allocated blocks as needed. So you won't "run out" of space. The EQL GUI just won't match what the OS says.
Regards,
Don
Suyash83
4 Posts
0
May 29th, 2018 08:00
Thank you Don for response.
Is their any option to release space or stop auto growing volume space.
dwilliam62
1 Rookie
1 Rookie
•
1.5K Posts
0
May 29th, 2018 13:00
Hello,
Not from the array side. The array is not aware of files, filesystems or allocation tables. The Server sends a SCSI WIRTE command to a specific Logical Block Address (LBA). If you delete a file using that LBA, the host OS allocation table marks it as free. But nothing sent to the storage to say it's free. That's where the UNMAP command comes in. When the OS frees that LBA (or LBAs) it is responsible for sending the UNMAP command(s).
However, XenServer doesn't 'own' the filesystem on that LUN. It's simply sliced up for use by the VMs. They don't own the volume so they can send the UNMAP command either.
So unless you have enough space to create a new volume, migrate all the VMs to that, then delete the old volume you can get the space back with XenServer.
Otherwise you would have to connect to your Data volumes from inside the guest. Know as "Storage direct" or "guest initiated iSCSI". The if the VM OS supports UNMAP those data volumes will be trimmed.
However, if the volumes use EQL replication, sync or async, then UNMAP is not supported at all.
Regards,
Don
Suyash83
4 Posts
0
May 30th, 2018 14:00
Thanks Don for response.
dwilliam62
1 Rookie
1 Rookie
•
1.5K Posts
0
May 31st, 2018 08:00
Hello Syash83,
You are very welcome. I am glad that I could assist you.
Don
Suyash83
4 Posts
0
June 1st, 2018 07:00
Hi Don,
I run reclaimed free space command and good news is, Citrix volume released more then 1 TB space. ( eariler its was 6.23TB used and now it is showing 4.14TB used.
I have allocated 1.4TB space for Citrix volume snapshot. now I observing snapshot size is auto growing. ( I have already disabled snapshot shedule)
need your inputs.
Regards,
Suyash
dwilliam62
1 Rookie
1 Rookie
•
1.5K Posts
0
June 1st, 2018 14:00
Hello,
If you have at least one snapshot then as writes occurs to the base volume the snapshot usage will increase. It's how all snapshots work. You have to have a copy of the old data to go back to.
If you don't want snapshots then set the reserve to zero, or greatly reduce the snapshot reserve size. Thin provisioning the volume will also thin provision snapshot reserve.
Also snapshot can impact reclaim operations if the page to be released is part of a snapshot it won't be returned to free space. In case you need to restore the data that was there originally.
Regards,
Don