Yes, I’ve made sure its not snapshots, but willing to change my workflow thats causing this issue in the first place.
Sanity check on the use case first (and I’ll take recommendations for doing things differently if it makes more sense):
tdarr temp drive location (/mnt/temp), to avoid reading and writing a lot to the main pool with the movies (/mnt/tank). temp drive is just a single 640GB drive, main pool is 5x8Tb in RaidZ1. tdarr takes a copy of the media file, moves it to temp location. Usually, tdarr is for transcoding which is a longer process, but I’m mostly just making lots of smaller changes, audio, metadata changes, etc. So the file turnaround is pretty quick for the edits to be made and the output file transferred to the tank pool to replace the original unedited file, and the in-progress work file gets deleted. Now keep in mind I’m doing this for my whole movies and tv library, up to nearly 10,000 files, with about 8 transcode threads going on at once, though I think I had errors with less still.
After letting tdarr run for a bit, I start having errors about the transcode files, that they don’t exist after trying to be used after copying. Seems like the drive is nearly full, according to the TrueNas Datasets view. Yet the du command shows its empty.
root@SHFN[/mnt/temp]# zfs list -r -t filesystem,volume -o space temp
NAME AVAIL USED USEDSNAP USEDDS USEDREFRESERV USEDCHILD
temp 27.5G 546G 0B 96K 0B 546G
temp/tdarr 27.5G 546G 0B 546G 0B 0B
root@SHFN[/mnt/temp]# du -h -d 1 /mnt/temp
512 /mnt/temp/tdarr
1.0K /mnt/temp
root@SHFN[/mnt/temp]# zfs get compression,compressratio,usedsnap,usedds,referenced,refquota,refreservation,logicalused,logicalreferenced,quota,version temp
NAME PROPERTY VALUE SOURCE
temp compression lz4 local
temp compressratio 1.00x -
temp usedbysnapshots 0B -
temp usedbydataset 96K -
temp referenced 96K -
temp refquota none default
temp refreservation none default
temp logicalused 547G -
temp logicalreferenced 42K -
temp quota none default
temp version 5 -
I know I’ve seen people say 'TrueNas takes awhile to delete files" but its been hours and I’m not seeing any progress in that whole time. It fixed itself once before but wasn’t exactly sure what, because I think it stayed like that immediately after a reboot, but maybe it started going again after that?
Am I hitting a bug here? or perhaps I’m really not using zfs in the way its meant to be with my workflow? Thanks for reading this. Using Truenas Scale 25.04.3