The attached patch provides a solution which ensures the return in the right condition and after the stat collection. The code structure requires another ugly boolean to proceed so as to preserve doing the reconnection between the decision that the return must be done and the place where it can be done, after reconnecting.


Ooops, the attached patched was the right content but wrongly named:-(

Here it is again with a consistent name.

Sorry for the noise.

--
Fabien.
diff --git a/src/bin/pgbench/pgbench.c b/src/bin/pgbench/pgbench.c
index 87fb006..8c5df14 100644
--- a/src/bin/pgbench/pgbench.c
+++ b/src/bin/pgbench/pgbench.c
@@ -1766,7 +1766,8 @@ doCustom(TState *thread, CState *st, StatsData *agg)
 {
 	PGresult   *res;
 	Command   **commands;
-	bool		trans_needs_throttle = false;
+	bool		trans_needs_throttle = false,
+				return_before_next_trans = false;
 	instr_time	now;
 
 	/*
@@ -1849,6 +1850,8 @@ top:
 
 	if (st->listen)
 	{							/* are we receiver? */
+		bool listened_a_meta = commands[st->state]->type == META_COMMAND;
+
 		if (commands[st->state]->type == SQL_COMMAND)
 		{
 			if (debug)
@@ -1892,6 +1895,7 @@ top:
 			/*
 			 * Read and discard the query result; note this is not included in
 			 * the statement latency numbers.
+			 * Should this be done before recording the statement stats?
 			 */
 			res = PQgetResult(st->con);
 			switch (PQresultStatus(res))
@@ -1913,6 +1917,7 @@ top:
 		{
 			if (is_connect)
 			{
+				/* Should transaction stats recorded above count this time? */
 				PQfinish(st->con);
 				st->con = NULL;
 			}
@@ -1942,12 +1947,17 @@ top:
 			 * listen back to true.
 			 */
 			st->listen = false;
+
+			if (listened_a_meta)
+				return_before_next_trans = true;
+
 			trans_needs_throttle = (throttle_delay > 0);
 		}
 	}
 
 	if (st->con == NULL)
 	{
+		/* Why is connection time is out of transaction time stats? */
 		instr_time	start,
 					end;
 
@@ -1969,6 +1979,10 @@ top:
 		memset(st->prepared, 0, sizeof(st->prepared));
 	}
 
+	/* ensure that meta-only scripts sometimes return */
+	if (return_before_next_trans)
+		return true;
+
 	/*
 	 * This ensures that a throttling delay is inserted before proceeding with
 	 * sql commands, after the first transaction. The first transaction
-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Reply via email to