Sqlite timeout settings does not appear to have effect sometimes

[morsok]

Sqlite timeout settings does not appear to have effect sometimes

[gotson]

There's 2 different timeouts:

• the connection pool timeout. Komga uses a connection pool, with a default of 1 connection. If multiple threads want to access the DB at the same time, they need to wait for a connection to be available from the pool.
• the sqlite busy_handler timeout. When a DB connection is trying to access the DB while another connection is locking the DB (in case of a write), the connection will wait up to that timeout to retry.

What you are experiencing is a connection pool timeout, which is set at 30s and cannot be changed by configuration as of now.

if you run Komga on low-end hardware, i would suggest that you:

• lower the number of task threads to a minimum like 1 (by default it is 8, which is probably too high for some cases)
• set the pool-size to 1 more than the number of task threads (so 2). This will leave some room to breathe for the API when requesting data.

[morsok]

I've followed your recommendation concerning the settings sadly I still have timeouts, the NAS is not the best hardware for speed.
Feel free to close this issue if you don't want to expose the second settings in configuration, not much can be done here.

[gotson]

Exposing the pool timeout would probably not lead to any improvement, it would just delay the inevitable.

I'd prefer trying to attempt fixing the bottlenecks, some queries might need optimization.

Could you share a log file so I can maybe spot where the issue happens?

Can you also provide some info on how many items are in the db (metrics tab would show that), and the db size in gb.