lock_multi unit test fails

[Lee Bieber]

Both Jay and I are seeing this failure. I'm seeing my failure on when running on Mac and Ubuntu

main.lock_multi [ fail ]

--- /home/lbieber/bzrwork/drizzle-clean-code/tests/r/lock_multi.result 2009-01-29 18:22:06.000000000 +0300
+++ /home/lbieber/bzrwork/drizzle-clean-code/tests/r/lock_multi.reject 2009-01-30 06:47:00.000000000 +0300
@@ -6,7 +6,7 @@
 select n from t1;
 unlock tables;
 n
-1
+4
 drop table t1;
 create table t1(n int);
 insert into t1 values (1);

drizzletest: Result content mismatch

[Jay Pipes (jaypipes)]

This occurs only periodically for me, not every time the regression suite is run. So, I am going to keep an eye on it.

[Padraig O'Sullivan (posulliv)]

I ran this test case 10 times and it failed once. So the behavior I get is that it passes the majority of times but fails once in a while.

[Jay Pipes (jaypipes)]

Yep, that's about the frequency I see it, too... definitely need to keep an eye on this. Unfortunately, when I see it fail, it's very tough to reproduce it right afterwards, and impossible within GDB. :(