Skip to content

Conversation

tcely
Copy link
Contributor

@tcely tcely commented Apr 8, 2025

The accumulation of memory can be mitigated by ending the queue workers periodically.

s6-supervise is expected to launch a new process afterwards.

Adding an offset to the sleep should help with SQLite concurrency issues. By starting tasks at different times, the chances of finding the database is locked should be reduced.

Change the settings for background_task workers without additional user intervention.

Fixes #428
Fixes #927
Fixes #955

@tcely tcely marked this pull request as ready for review April 8, 2025 18:23
@tcely tcely moved this to In Progress in Status Apr 8, 2025
@tcely tcely moved this from In Progress to Ready in Status Apr 8, 2025
tcely added 7 commits April 9, 2025 02:12
Move thumbnail download from NET to FS.
It does use the network, but not with `yt_dlp` and not in a way that YouTube cares about.
The parser change and a few other things mean that compatibility with `3.8` is unlikely to last much longer.
@tcely tcely mentioned this pull request Apr 11, 2025
@tcely tcely mentioned this pull request Apr 12, 2025
tcely added 5 commits April 13, 2025 04:32
The `sqlite` driver appears to be fundamentally incompatible with the Django query set `iterator` function.

It only works for the first chunk, then it fails any further database operations because the database is already locked.
@meeb
Copy link
Owner

meeb commented Apr 13, 2025

@tcely This has conflicts. Thanks!

@tcely
Copy link
Contributor Author

tcely commented Apr 13, 2025

@meeb The merge conflicts have been resolved.

@meeb meeb merged commit 8c729d0 into meeb:main Apr 13, 2025
7 checks passed
@github-project-automation github-project-automation bot moved this from Ready to Done in Status Apr 13, 2025
@tcely tcely deleted the patch-2 branch April 13, 2025 12:34
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Archived in project
Development

Successfully merging this pull request may close these issues.

High Memory Usage Tubesync limited to single task after 2025-04-06 update Huge memory use
2 participants