db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Myrna van Lunteren (JIRA)" <derby-...@db.apache.org>
Subject [jira] Created: (DERBY-625) hang in unit/cacheService.unit with wctme5.7
Date Thu, 13 Oct 2005 18:17:05 GMT
hang in unit/cacheService.unit with wctme5.7 

         Key: DERBY-625
         URL: http://issues.apache.org/jira/browse/DERBY-625
     Project: Derby
        Type: Bug
 Environment: windows with wctme5.7 (j9 version 2.2)
    Reporter: Myrna van Lunteren

Test unit/cacheService.unit hangs with wctme5.7 jvm since June...

Can someone familiar with services please have a look?

The hang was not noticed in a timely fashion because our scripts only run with wctme5.7 once
a week, 
and for 3 months, our nightly tests took too long to complete - and no one paid attention
(me included).

I've done some historical research on this, but am no closer to deciding if this is a jvm
bug exposed by changed derby functionality or a 'new' derby bug exposed by the wctme5.7 (j9
version 2.2) jvm.
Efforts are hampered by 
- absence of commit archive details between June 13 and July 1
- svn cannot backtrack to revisions that happened during incubator time

This is what I found so far:
- in main
	- the test ran fine on 6/18/2005, but failed on 6/25/2005.
	- From saved commits emails, it appears relevant revisions are between  191353 (jean)
          and 201792 (kathey), jean's are all dita related,  2 of kathey's are functionTests
only changes.
	  This leaves the following 2
		201792 - kmarsden - DERBY-389,
		191755 - kmarsden - DERBY-8 and DERBY-366; mostly test/master updates, but additionally
	- I backtracked the changes from these revisions but still the hang persists...
- 10.1 
	- I tested build 224373 and 208786 and the could reproduce the hang; but 
	  with build 190628 (which happened to be stting around on my machine) there was no hang.
	  Thus, with 10.1 branch, the problem was introduced/backported between revision
	  190628 and revision 208786.
	 - Looking at a combination of the commit archive for July and my saved commit emails, again,
most of these are innocent test or dita updates.
	 - The following are more interesting:
		- 201659 - satheesh - DERBY-388 - 
		- 208697 - dan - cleanup of published javadoc for embedded jdbc objects
		- 208776 - andrew - monster commit backport fixes for DERBY-8, DERBY-247, DERBY-276, 
		  DERBY-348, DERBY-354,DERBY-361, DERBY-366, DERBY-389, DERBY-405, DERBY-416, and DERBY-420
 	- I tried to backtrack my 10.1 branch to revision 208775, but that resulted in:
		svn: REPORT request failed on '/repos/asf/!svn/vcc/default'
		Cannot replace a directory from within

In short, historical research has gotten me nowhere.
I looked at the stack of the threads when the program is hanging but it tells me nothing either.

I will attached the stack threads obtained by running the following command:
  j9 -jcl:max -Xbootclasspath/p:c:/ive22/lib/database_enabler.jar;c:ive22/lib/classes.zip
 on windows and subsequently pressing ctrl-break, and derby.log

Note: this problem does not occur with wsdd5.6 (j9 version 2.1). The test never runs/ran with
j9foundation (also wctme5.7).	

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators:
For more information on JIRA, see:

View raw message