"Paul" == Paul Letta via toasters toasters@lists.teaparty.net writes:
From: Paul Letta letta@jlab.org To: toasters toasters@teaparty.net Date: Tue, 15 Jul 2025 20:57:00 +0000 Subject: Weird mass delete of files in a volume with ONTAP ASYNC-Delete
Hi Toasters.
I’m running 9.14.1P13.
I used the ONTAP ASYNC-Delete feature from the GUI (storage – Volumes – click on a volume – Click on File system – click on Explorer) to delete a few top level folders in a volume. That worked fine. I saw the background job running and they finished fine.
I needed to delete much more. So I clicked on the trash can for one of the folders, clicked ok to start the delete.
Why not do the delete from a CIFS share instead?
Then I clicked elsewhere in the GUI and clicked back to File-System, Explorer – and clicked on the trash can for a different folder.
Hmm... I wonder if you left something selected elsewhere, or even up a level when you started one of the deletes?
I did this maybe 6 or 7 times. When I check the jobs in ONTAP, I saw them all running.
But then as I waited, I saw a lot more space being used by snapshots (via the df command) than the 6 or 7 folders I wanted to delete.
At this point, only 1 ASYNC-Delete job was still running, so I canceled it.
When I look at the live file system.. I see almost all folders (not just the 6 or 7 I clicked) have massive files deleted. Luckily for me, this volume was just a filesystem that barely changes.. so I could snapshot restore from a few hours before, and all was well.
Whew! Love snapshots.
Now I’m scared to death of that file-explore page in the ONTAP GUI. I am the only administrator, so no one else could have been logged on to ONTAP. The pattern of deleted files makes a delete coming from a CIFS client very unlikely. There is no NFS export to this volume.
I've never used it, but I'd probably not ever use it unless I had a problem doing the deletes from a client instead. But I'm much more in tune with NFS than CIFS stuff.
Maybe the file explorer delete button was not intended for multiple, simultaneous runs?
Do you have the space to clone the volume to a test copy and play around and see if you can recreate the problem?
I don’t know.. I know I will avoid page like a plague forever now. Any thoughts ?
Do all filesystem manipulations from the clients if at all possible in the future.
John