We are getting problems with driver.ErrBadConn, but only if we use the prepared statements. It seems to me that this is because DB.QueryContext retries with a fresh connection and Stmt.QueryContext does not. I want to verify whether this is a bug or an expected behaviour,
When DB.QueryContext receives driver.ErrBaddConn from the driver it will retry twice (maxBadConnRetries) using cached connection (cachedOrNewConn strategy). If this fails, it will retry once more using new connection (alwaysNewConn strategy). The code is here: https://golang.org/src/database/sql/sql.go?s=7997:9207#L1223 Stmt.QueryContext will also retry twice (again maxBadConnRetries) using the cached connection. If this fails, it will return driver.ErrBadConn, without retrying using the new connection. The code is here: https://golang.org/src/database/sql/sql.go?s=7997:9207#L1935 s.connStmt uses cachedOrNewConn strategy. I think this is a bug (but I will be happy to hear otherwise), introduced in this commit: https://github.com/golang/go/commit/c468f94672af25bc34975ba96309e20e972fa340 Before this commit, maxBadConnRetries was 10 and retries where alway done using old (cached) connections. Commit changed the logic, decreasing the number of retires and adding one more retry on fresh connection. While decrease was global (both *DB and*Stmt), retry-using-fresh-connection logic was applied only to *DB. So is it a bug (aka something I can report on https://github.com/golang/go/issues) or is it an expected behaviour? Thanks, Krzysztof Dryś -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
