Bug #17359 Test suite starts server before determining whether test should be run
Submitted: 14 Feb 2006 0:32 Modified: 20 May 2006 21:04
Reporter: Greg Lehey Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Server: Compiling Severity:S3 (Non-critical)
Version:5.0 OS:Not specific
Assigned to: Kent Boortz CPU Architecture:Any

[14 Feb 2006 0:32] Greg Lehey 
Description:
It appears that the test suite starts the server before determining whether an individual test should be run.  This causes problems in the case where the server start options are not recognized.  The server crashes (correctly) and the test never gets as far as determining that it shouldn't be run.

This occurred during the solution of BUG#14286.  This bug requires building a server without support for transactional storage engines.  The test innodb_unsafe_binlog should clearly not be run, but the test suite attempts to start the server with the option --innodb_locks_unsafe_for_binlog=true.  As built, the server does not recognize this option, and it stops.

A related issue is that the test suite does not notice that the server has stopped: instead, it waits until the client connection times out.  The PERL version of the tests does not suffer from this problem (it reports the server crash immediately, which it still treats as an error).

How to repeat:
./configure --without-innodb --without-berkeley-db --without-ndbcluster
make
make test
[14 Feb 2006 0:40] 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/2559
[14 Feb 2006 14:17] Elliot Murphy 
Patch was for bug#14286, not this bug.
[21 Apr 2006 18:25] 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/5317