Best recording of a species gets deleted

[krummrey]

PLEASE READ THE DOCUMENTATION BEFORE SUBMITTING AN ISSUE

Describe the bug
Recordings are deleted when disk space is low. the script does not take into account if the recording that is being deleted is "the best recording" of a species. When I browse through the list of species detected i have a lot of entries where no recording is available any more.

To Reproduce
Check Best recordings. Mine has many with no spectogram or recordings for some birds available

Expected behavior
Recordings and spectogram should not be removed if they are the best the script ever recorded for that species

Screenshots

Additional context

Your build
Raspi 4 with 4 gig of RAM

Code or log snippets

[alexbelgium]

Hi, I had seen the same and initially proposed a PR : #276 I'll reactivate it

Instead of purging indescriminately, the PR uses the same logic as disk species clean by removing first files with lower confidence

[Nachtzuster]

Hmm, first I thought this was just a limitation of the disk_check.sh script, but it turns out that scripts actually triggers the creation of scripts/disk_check_exclude.txt which has the list of the best recording for each species.

That seems to be working as it should on my test setup when I run the ./scripts/disk_check.sh manually (ie best recording is kept).

Could you do

ls -l scripts/disk_check_exclude.txt
cat scripts/disk_check_exclude.txt

Also could you go into the 'Recordings' page, pick some bird, sort by confidence and check if the first (ie best) recording is locked?