MySQL Bugs: #39484: main.concurrent_innodb

Bug #39484	main.concurrent_innodb_safelog fails sporadically
Submitted:	16 Sep 2008 15:38	Modified:	14 Jan 2013 10:11
Reporter:	Alexander Nozdrin	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Tests: Server	Severity:	S3 (Non-critical)
Version:	6.0-TRUNK,5.4	OS:	Any
Assigned to:	Assigned Account	CPU Architecture:	Any
Tags:	pushbuild, sporadic, test failure

Description:
main.concurrent_innodb_safelog fails sporadically as follows:

@@ -72,7 +72,7 @@
 ** Release the lock and collect result from update on thread 2
 select release_lock("hello");
 release_lock("hello")
-1
+NULL
 ** Table should have eta updates where tipo=11 but updates made by
 ** thread 1 shouldn't be visible yet.
 select * from t1;

The bug reported after the failure in mysql-6.0 on Mon Sep 15 20:48:13 2008
(http://tinyurl.com/6exj73) on sapsrv2.

How to repeat:
http://tinyurl.com/5qzrkh

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/66304

2754 Horst Hunger	2009-02-13
      Fix for bug#39484: I increased the get_lock timeout to 300 seconds as 10 second are too short for machines with low performance and a release_lock may end up with NULL due to a failed get_lock. I also replaced the sleeps by wait condition.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/66717

2838 Tomas Ulin	2009-02-17
      disabled test: Bug #39484 main.concurrent_innodb_safelog fails sporadically
      modified:
        mysql-test/t/disabled.def

Pushed into 5.1.32-ndb-6.2.17 (revid:tomas.ulin@sun.com-20090218084425-guqwr1xa34ctwi4j) (version source revid:tomas.ulin@sun.com-20090217173919-31hlyng1rnpu05b8) (merge vers: 5.1.32-ndb-6.2.17) (pib:6)

Pushed into 5.1.32-ndb-6.3.23 (revid:tomas.ulin@sun.com-20090218084629-1abu96q2a8ii8fac) (version source revid:tomas.ulin@sun.com-20090218084629-1abu96q2a8ii8fac) (merge vers: 5.1.32-ndb-6.3.23) (pib:6)

Pushed into 5.1.32-ndb-6.4.3 (revid:tomas.ulin@sun.com-20090218084745-9oypy4i4nzplzv6j) (version source revid:tomas.ulin@sun.com-20090218084745-9oypy4i4nzplzv6j) (merge vers: 5.1.32-ndb-6.4.3) (pib:6)

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/66986

2754 Horst Hunger	2009-02-20
            2. fix for bug#39484: I resetted the timeout values of get_lock as they have no influence on the actual bug. Due to bug 32782 I added NULL as value to the release_lock as being successful.
            I also replaced the first 2 "sleep 1" by wait conditions. the last 2 sleeps have not been replaced as I have not found a proper condition leading to deterministic results. All is described in comments.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/67153

2754 Horst Hunger	2009-02-23
      2. fix for bug#39484: I resetted the timeout values of get_lock as they have no influence on the actual bug. Due to bug 32782 I replaced "SELECT release_lock" by "DO release_lock" to catch the varying results as being successful.
                  I also replaced the first 2 "sleep 1" by wait conditions. the last 2 sleeps have not been replaced as I have not found a proper condition leading to deterministic results. All is described in comments.

Results of the review:
1. I also have no idea how to replace the remaining sleeps
2. Formal stuff:
   1. Line 14 "Hallo" -> "Hello"
   2. Line 120,130,... 
      "proably" -> "probably"
      "bug 32782" -> "Bug#32782 User lock hash fails to find lock"
      "bug#39484" -> "Bug#39484 main.concurrent_innodb_safelog
                                fails sporadically"
      The main reasons (bug title is here sufficient) must
      be mentioned.
      A reader should find all important information within
      the test and not be forced to look into the Bugs DB.
      IMHO a unification of "bug writing style" is strongly
      recommended. 
      ->   Bug#nnnnn <Title>
      If this is not foillowed tests look chaotic and ugly.
   3. Line 94, ...
      Please remove the trailing spaces
3. The
   DO release_lock("hello");
   DO release_lock("hello2"); 
   at begin and end of test do not have the intended effect
   because the manual says
   RELEASE_LOCK(str)
   Releases the lock named by the string str that was
   obtained with GET_LOCK().
   Returns 1 if the lock was released,
   0 if the lock was not established by this thread
   (in which case the lock is not released).

   Protocol of my experiment:
   DO release_lock("hello");
   CREATE USER mysqltest@localhost;
   "# Establish connection thread1 (user=mysqltest)
   SELECT get_lock("hello",3);
   get_lock("hello",3)
   1
   "# Switch to connection default
   SELECT release_lock("hello");
   release_lock("hello")
   0     <--------- !!!
   SELECT get_lock("hello",3);
   get_lock("hello",3)
   0     < -------- !!!

   Please remove them all.

Fixed in 5.6.