How to properly debug minflux if 'LOG_LEVEL=debug' is not enough?

[rwiesler]

Hi,

I need some hints because I did run regulary into timout errors when logging into miniflux with (> 60 seconds) and needed multiple relogins until the unread page has been fully build.

As far I can tell it seems to be related to "Setttings/Entries per page (Default: 100)" used as parameter for limiting the SQL query.
Increasing the limit and activating "Hide entries in global unread list" does the trick for now but I like to digg a little deeper.

There are roughly over 800 feeds and over 530.000 unread entries (in total over 700.000) in the database, but I like to take a direct look at the timers within the page to conform my suspicion.

v2/internal/ui/unread_entries.go

Line 28 in 27253c8

beginSqlCountUnreadEntries := time.Now()

It clearly also affects 'History', 'Feeds' and 'Category'.

And looking at at the hardcoded SQL limit at the API endpoint

v2/internal/api/entry.go

Line 108 in 27253c8

limit := request.QueryIntParam(r, "limit", 100)

this varible might be better a shown and made editable in the settings for the API Integrations?

Site note: EXPLAIN ANALYZE of the query build by the ORM with LIMIT 100 and without LIMIT 100 is clearly in favour without the LIMIT 100 in loading times.
Only as a rought idea: loading everthing at once seems better would then require to limit the buildup of the HTML DOM with vanilla ECMAScript.

Abonded project as a proof of concept for limiting the entries per page within the HTML DOM.
https://github.com/Mobius1/Vanilla-DataTables

Thanks.