I have been grinding on for quite some time on why "now" any query (or model refresh) will time-out after 10 mins, being very certain a query could run for longer in the "past".
The only likely candidate I can find is this commit
Unless someone can tell me I am reading the context of this code all wrong, or come forward and show they are running queries for longer against a PostgreSQL engine.
New year, new possibilities.
Would it be possible to remove this hardcoded and arbitary limit?
As noted in the pr, a statement timeout may be set in the connection string if it is needed.
as you can imagine, we put that limit in our code because of a reason: we lock threads when doing so and we have people that save hundreds of models. I would suggest you fork the project, remove that line and have a Metabase tailored to your needs
Your reasoning does not make sense. The transaction is clearly per query/model, so the number of models (we have 52 atm.) does not factor in. There may be longer running transactions related to sync, finger printing, or what ever; but why is 10 mins a magic number?