Hello Tomas,
So I guess this is a bug in 12788ae49e1933f463bc59a6efe46c4a01701b76, or
one of the other commits touching this part of the code.
I could not reproduce this issue on head, but I confirm on 11.2.
AFAICS on head it's fixed by 3bac77c48f166b9024a5ead984df73347466ae12
Thanks for the information.
I pinpointed the exact issue in one go: no surprise given that the patch
was motivated by cleaning up this kind of external state machine changes
which I thought doubtful and error prone.
Attached is a fix to apply on pg11.
--
Fabien.
diff --git a/src/bin/pgbench/pgbench.c b/src/bin/pgbench/pgbench.c
index 9eaa192239..e2093cdaca 100644
--- a/src/bin/pgbench/pgbench.c
+++ b/src/bin/pgbench/pgbench.c
@@ -5762,6 +5762,10 @@ threadRun(void *arg)
}
}
+ /* under throttling we may have finished the last client above */
+ if (remains == 0)
+ break;
+
/* also wake up to print the next progress report on time */
if (progress && min_usec > 0 && thread->tid == 0)
{
