1 Rookie
•
5 Posts
•
2 Points
0
60
March 23rd, 2026 03:40
Powerflex R3_6.1 small amount of free space compared with sum of volumes and snapshots
Hello Guys,
We have a situation where it seems that we have some invisible bloat in a powerflex standalone cluster . The cluster consists of 13 Dell servers dedicated to powerflex only , with combined capacity of 157 TB . Which is validated by scli printout :
Physical data:
157.0 TB (160813 GB) total capacity
There is 34% of spare capacity , again lining up with the printout :
53.4 TB (54676 GB) spare capacity
And none of the Disks are faulty or out of service , or have any errors showing :
0 Bytes decreased capacity
All of the SDSs are online and healthy . Currently the Major capacity alarm is triggered , which is set for 90% .
However total sum of all the volumes created in the system of which there are 50 , all of them thick provisioned and 61 snapshots is approx 22 TB ,
This cluster is set up for RAID 1 , so in a worst case scenario there should be , 157 - 54 - (2*22) = approx 59 TB of RAID1 space or approx 29.5 TB of volume space . Yet we have hit 90% alarm , and if we create another test volume we can trigger the Critical Alarm .
So the system definitely believes that we are currently at 90% used capacity , but we can not account for it .
It looks as if there are stale allocations from old deleted volumes in the system .
Is there a way of looking up the allocations per volume , so that we can validate the usage ?
Or any other way to troubleshoot the phantom used space ?
But most importantly is there a way to flush out this bloat ?
Any thought or advice or other information would be very welcome .
0 events found
DELL-Josh Cr
Moderator
•
9.6K Posts
•
110 Points
0
March 25th, 2026 12:59
Once a VTree grows, it does not automatically shrink just because:
volumes are deleted
snapshots are deleted
only a tiny leftover object remains
At some earlier point:
A very large thick‑provisioned volume (or multiple) existed in that VTree
Or large snapshots were created off it
Or both
Because the VTree is thick‑provisioned, PowerFlex pre‑allocates all extents upfront.
That immediately grows the VTree to its maximum historical size.
The base volume was deleted
Almost all snapshots were deleted
One tiny leftover snapshot remains:
PowerFlex will never shrink a VTree automatically, because
VTrees are designed for snapshot chains
Automatic shrinking could invalidate snapshot metadata
The system assumes the VTree may grow again
These commands should delete the vtree.
scli --query_all_vtrees
scli --delete_volume --volume_id <snapshot_id>
scli --query_vtree --vtree_id <id>
DELL-Josh Cr
Moderator
•
9.6K Posts
•
110 Points
0
March 23rd, 2026 12:52
Hi.
Thank you for your question.
Snapshots allocate their full size when they are created.
scli --query_vtree --volume_name <volume_name>
look at User data without trimmed capacity
Most likely deleting old snapshots will be the fix.
Let us know if there is anything else we can assist you with.
Thanks,
Josh
Did I answer your query? Please click on ‘Mark as Accepted Answer’. ‘Thumbs up’ the posts you like!
#Iwork4Dell
mil4898
1 Rookie
•
5 Posts
•
2 Points
0
March 24th, 2026 00:14
@DELL-Josh Cr
Hello Much thanks for the information . I have had a look at the vtrees already , and we have 5 of them in total . Here is a printout of their footprint :
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 1.4 TB (1424 GB)
Total user data: 1.4 TB (1424 GB)
Total base user data: 640.0 GB (655360 MB)
Total snapshots user data: 784.2 GB (802979 MB)
Number of Volumes: 26
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 21.0 GB (21490 MB)
Total user data: 21.0 GB (21490 MB)
Total base user data: 16.0 GB (16384 MB)
Total snapshots user data: 5.0 GB (5106 MB)
Number of Volumes: 2
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 350.0 GB (358391 MB)
Total user data: 350.0 GB (358391 MB)
Total base user data: 104.0 GB (106496 MB)
Total snapshots user data: 246.0 GB (251895 MB)
Number of Volumes: 12
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 277.0 GB (283643 MB)
Total user data: 277.0 GB (283643 MB)
Total base user data: 200.0 GB (204800 MB)
Total snapshots user data: 77.0 GB (78843 MB)
Number of Volumes: 3
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 561.7 GB (575152 MB)
Total user data: 561.7 GB (575152 MB)
Total base user data: 448.0 GB (458752 MB)
Total snapshots user data: 113.7 GB (116400 MB)
Number of Volumes: 12
They are smallish , not much usage there 2 to 3 TB, when we compared it to the 29.5 TB that we should have , but we can not find .
Best Regards
(edited)
mil4898
1 Rookie
•
5 Posts
•
2 Points
0
March 24th, 2026 06:14
@DELL-Josh Cr
Disregard that post , looks like we have more than that . I will go through them all to see how they add up . To avoid confusion lets delete my follow up post completely .
mil4898
1 Rookie
•
5 Posts
•
2 Points
0
March 25th, 2026 01:18
Hello Guys ,
Update
In fact we had 54 Vtrees , not 5 as originally thought .
And yes one of the Vtrees was bloated to the max :
> scli --query_vtree --vtree_id xxxxxxxxxxxxxxx
VTree ID: xxxxxxxxxxxxxxxxx Name: xxxxxxxxxxxxxxxxxxxx
Storage Pool xxxxxxxxxxxxxx Name: xxxxxxxx
Protection Domain xxxxxxxxxxxxx Name: xxxxxxx
Data layout: Medium granularity
Provisioning: Thick
Total capacity in use: 38.0 TB (38912 GB)
Total user data: 38.0 TB (38912 GB)
Total base user data: 38.0 TB (38912 GB)
Total snapshots user data: 0 Bytes
Migration status: Not in migration
?> Volume ID: xxxxxxxxxxxxx Name: xxxxxxxxxxxxxxxxxxx
Snapshot of a deleted volume
Size: 16.0 GB (16384 MB)
Creation time: 2025-07-31 08:46:06
Volume is not mapped
No SCSI reservations
Doesn't use RAM Read Cache
Replication info:
Replication state: Not being Replicated
This Vtree was 38TB , and yet only had one volume of 16 GB associated with it .
How is that possible ? How does a Vtree grow so big ?
mil4898
1 Rookie
•
5 Posts
•
2 Points
0
March 25th, 2026 23:35
Much thanks for that , appreciate the help . The volume was deleted and with it the Vtree has gone . And all of the held space has now been released . The alarm has cleared and we are back to normal .
(edited)
DELL-Josh Cr
Moderator
•
9.6K Posts
•
110 Points
0
March 26th, 2026 11:40
Great to hear.