What causes "strings.Join(failures, "; ")" in t8y2/dbx?

Inspect the joined message(s) to see which cleanup step failed; a 'closed' style message is usually safe to ignore

strings.Join(failures, "; ")

Error message

strings.Join(failures, "; ")

What it means

disconnect tears down the session's database and connection and aggregates every teardown error into a single error whose message is the failure strings joined with "; ". If any Close() (or related cleanup) fails, disconnect returns that combined error so the caller knows cleanup was only partial. The bracketed message is the joined-error text itself, e.g. "connection already closed".

Source

Thrown at agents/drivers/argo-go/main.go:530

	server.connectionMu.Lock()
	connection := server.connection
	database := server.database
	server.connection = nil
	server.database = nil
	server.connectionMu.Unlock()
	var failures []string
	if connection != nil {
		if err := connection.Close(); err != nil {
			failures = append(failures, err.Error())
		}
	}
	if database != nil {
		if err := database.Close(); err != nil {
			failures = append(failures, err.Error())
		}
	}
	if len(failures) > 0 {
		return errors.New(strings.Join(failures, "; "))
	}
	return nil
}

func (server *server) requireConnection() (*sql.Conn, error) {
	server.connectionMu.Lock()
	connection := server.connection
	server.connectionMu.Unlock()
	if connection == nil {
		return nil, errors.New("Hive connection is not open")
	}
	return connection, nil
}

func (server *server) setActiveOperation(cancel context.CancelFunc) {
	server.activeMu.Lock()
	server.activeCancel = cancel
	server.activeMu.Unlock()

View on GitHub (pinned to c0390bff16)

Solutions

Inspect the joined message(s) to see which cleanup step failed; a 'closed' style message is usually safe to ignore
Ensure disconnect is called exactly once per session (idempotent guards/once semantics)
Catch and log this error on shutdown paths rather than failing the whole shutdown
If the connection is already dead, force-reset server connection state instead of relying on graceful Close

Example fix

// before
err := server.disconnect() // err: "driver: bad connection"
if err != nil { return err } // aborts shutdown
// after
if err := server.disconnect(); err != nil {
    log.Printf("disconnect cleanup issues (ignoring): %v", err)
}

Defensive patterns

Strategy: try-catch

Try / catch

if err := server.disconnect(); err != nil {
    log.Printf("disconnect cleanup issues: %v", err) // joined with '; '
    // do not abort shutdown; inspect individual messages
}

Prevention

Call disconnect exactly once per session
Log rather than propagate teardown errors on shutdown paths
Reset connection state when the peer is already gone
Avoid concurrent disconnect calls with a mutex/once guard

When it happens

Trigger: Calling disconnect (directly or via openSession, dispatch, testConnection) when closing the *sql.DB/database or the sql.Conn fails — e.g. the network already dropped, the driver returns an error on Close, or both cleanup steps fail and messages are concatenated with "; ".

Common situations: Disconnecting after the HiveServer2 process restarted (socket already dead); TLS or auth state making the close handshake fail; double-disconnect from concurrent callers; flaky network causing partial cleanup.

Related errors

AI-assisted analysis of t8y2/dbx@c0390bff16 (2026-09-05). Data as JSON: /api/errors/65a908ab5819a3bf. Report an issue: GitHub.