The README now says "Since some middlewares close idle connections by 5 minutes,". Could you be more precise about which middlewares you are talking about?

https://golang.org/pkg/database/sql/#DB.SetMaxOpenConns
golang doc says

If MaxIdleConns is greater than 0 and the new MaxOpenConns is less than MaxIdleConns, then MaxIdleConns will be reduced to match the new MaxOpenConns limit.

so MaxIdleConns <= MaxOpenConns, why are we suggesting setting higher idle connection number here?

why are we suggesting setting higher idle connection number here?

I wrote "same to (or greater than)" because greater value become same value automatically.
If it is confusing, I'm OK to remove "(or greater than)" part.

@methane can you explain why you suggest to set max open connections equal to max idle connections?
Looks like sql.DB is still trying to make additional connections when there are idle connections (maybe due to burst of traffic), more details here: https://stackoverflow.com/questions/67653872/golang-sql-db-waitcount-greater-than-0-even-when-there-are-enough-idle-connectio

@methane can you explain why you suggest to set max open connections equal to max idle connections?

I explained it already. "When it is smaller than SetMaxOpenConns(), connections can be opened and closed very frequently than you expect."

Looks like sql.DB is still trying to make additional connections when there are idle connections (maybe due to burst of traffic), more details here: https://stackoverflow.com/questions/67653872/golang-sql-db-waitcount-greater-than-0-even-when-there-are-enough-idle-connectio

It looks like what you see is what I said. DB opens new connection because no idle connections. You didn't see high number of InUse just because number of InUse connections is moving like "spike", and your sampling rate is too slow to catch the spike.

You should try this as I recommended in the README.

db.SetMaxOpenConns(100)
db.SetMaxIdleConns(100)
db.SetConnMaxLifetime(time.Minute * 3)

If you can not believe me, you should get DB.Stats() every time before using DB methods. You will catch the spike.