Bug #50582 ndbmtd started on a single core machine causes Job Buffer Full

• Submitted: 25 Jan 2010 9:33
• Modified: 22 Feb 2010 8:58
• Reporter: Oli Sennhauser
• Status: Closed
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: mysql-5.1-telco-7.0
• OS: Any
• Assigned to: Andrew Hutchings
• CPU Architecture: Any
• Tags: 7.0.x, Contribution, job buffer full, ndbmt, single core

[25 Jan 2010 9:33] Oli Sennhauser

Description:
Customer experiences job buffer full with following node crash when running a ndbmtd on a single core machine.

How to repeat:
no idea.

Suggested fix:
Post a warning when this is detected in the cluster log for example.

[25 Jan 2010 23:44] Bugs System

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/98131

3363 Andrew Hutchings	2010-01-25
      Bug #50582  	ndbmtd started on a single core machine causes Job Buffer Full
      
      Add warning when MaxNoOfExecutionThreads > LockExecuteThreadToCPU count as follows:
      
      2010-01-25 23:35:52 [ndbd] WARNING  -- MaxNoOfExecutionThreads (4) > LockExecuteThreadToCPU count (2), this could cause contention

[26 Jan 2010 13:24] Magnus Blåudd

Since all the surrounding code uses the C++ interface to extract the value from iter I think this new code should do the same.

iter.get(CFG_DBXXX, &val);

[26 Jan 2010 16:06] Bugs System

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/98216

3363 Andrew Hutchings	2010-01-26
      Bug #50582  	ndbmtd started on a single core machine causes Job Buffer Full
      
      Add warning when MaxNoOfExecutionThreads > LockExecuteThreadToCPU count as follows:
      
      2010-01-25 23:35:52 [ndbd] WARNING  -- MaxNoOfExecutionThreads (4) > LockExecuteThreadToCPU count (2), this could cause contention

[2 Feb 2010 9:01] Bugs System

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/98898

[2 Feb 2010 9:08] Bugs System

Pushed into 5.1.41-ndb-7.0.13 (revid:[e-mail]-20100202090111-f2gvyfomndcjlvkr) (version source revid:[e-mail]-20100202090111-f2gvyfomndcjlvkr) (merge vers: 5.1.41-ndb-7.0.13) (pib:16)

[2 Feb 2010 9:18] Magnus Blåudd

Pushed to 7.0.12 and 7.1.2

[8 Feb 2010 14:20] Jon Stephens

Documented bugfix in the NDB-7.0.12 and 7.1.2 changelogs as follows:

        ndbmtd started on a single-core machine could sometimes fail
        with a Job Buffer Full error when MaxNoOfExecutionThreads was
        set greater than LockExecuteThreadToCPU. Now a warning is logged
        when this occurs.

Closed.