bug: timer_wait not working and db locked errors

[BaukeZwart]

Description of the bug

Running with this in my config, but when DL multiple episodes I see every episode is being processed by Plex the moment it's imported in Sonarr.
Also the log is filled with db locked errors. (Using SQLite)

opts:
  default_timer_wait: 180 # In seconds

triggers:
  sonarr:
    type: "sonarr"
    timer:
      wait: 5 # In seconds

ERROR autopulse::service::runner: unable to delete not found events: DatabaseError(Unknown, "database is locked")
ERROR autopulse::service::runner: unable to delete failed events: DatabaseError(Unknown, "database is locked")

Steps To Reproduce

See above

Additional Information

No response

[dan-online]

Hmm, this is odd as I recently added options to prevent this on SQLite

[BaukeZwart]

For now I reverted to autoscan but if there is anything I can test to help please let me know.
EDIT: I see that root is the owner of the db and the shm and wal files, is that correct?

[dan-online]

For now I reverted to autoscan but if there is anything I can test to help please let me know. EDIT: I see that root is the owner of the db and the shm and wal files, is that correct?

Yep that should be correct, and the presence of those files show it should be working correctly.. I'll do some testing on my end to see

[BaukeZwart]

Side note, I see you closed my earlier bug report about incorrect discord time stamp as solved.
But the issue is still there.

[dan-online]

@BaukeZwart the timestamp uses local time now, so as long as you set /etc/localtime:/etc/localtime:ro it should be correct, I'll add it to the docker-compose

[BaukeZwart]

I removed that from my compose when it didn't change anything during earlier testing.
Tried again and time stamp is correct now.

[dan-online]

Could you share more logs with log_level set to debug?

[BaukeZwart]

I have set log level to debug.
I will force a few imports later when my server is done with the bulk import from the seedbox and post the log.

[BaukeZwart]

Log: https://logs.notifiarr.com/?fa95c54a5d0df9e8#6rHLHTaU4B2mpSRQaFnJm7LEnENbsXMHM8k1cJD4R68m

[dan-online]

Hey @BaukeZwart can you test if #35 i.e. danonline/autopulse:pr-35 resolves your issue?

[BaukeZwart]

pulling it now, I will let you know.

[BaukeZwart]

@dan-online pr-35 seems to work much better.
I still see db locked errors in the log but the delay now seems to work.
This is the debug log after adding a 3 episode mini series.
https://logs.notifiarr.com/?586f50ab99ec1076#HK7i1jjcSU1Zc8M2c5Sny55bQKho4y7M6YUB9HVcYMew

[BaukeZwart]

[dan-online]

Hmm, I've increased the busy_timeout in the same tag, perhaps that might fix it?

[BaukeZwart]

That also didn't fix it.
I noticed the timestamp of the db itself never changes, it seems noting is written to it.

[dan-online]

That is extremely odd, could you try deleting the folder and then starting autopulse, seems to be some weird permissions issue

[BaukeZwart]

I tried that yesterday, stopped the container, deleted the db file and restarted.
I will try again but would be surprised if any changes.
EDIT: I checked, none of the other containers I run that use sqlite have the db with root as owner. They all use user: ${PUID}:${PGID} as the owner. Can that be the reason?

[dan-online]

That is very possibly the reason, I originally tried having the image not be root but that led to some permissions issues so I had to go root in #18

[BaukeZwart]

As a test I tried manually changing the perm and ownership of the db files and added ${PUID}:${PGID} to docker compose, but still the same db locked errors.

[dan-online]

Hmm, quite odd since I don't get these errors with example/docker-compose-sqlite.yml

[BaukeZwart]

I'm using the same docker compose.
Is there anything in the config that can cause this?

[BaukeZwart]

Switched to postgres and it all working fine now.

[dan-online]

I'll tentatively close this but if someone else sees this issue then feel free to comment to reopen this