[ 
https://issues.apache.org/jira/browse/DERBY-5599?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13201159#comment-13201159
 ] 

Kristian Waagan commented on DERBY-5599:
----------------------------------------

I verified that the istat feature isn't triggered on my system. Can you ckeck 
that it is running on the system where the test fails?
The easiest way to do do that is to add 
-DtestSpecialProps=derby.storage.indexStats.log=true^derby.storage.indexStats.debug.trace=true
 and run the test.
Note especially the line logged on shutdown:
Mon Feb 06 09:44:31 GMT+01:00 2012 Thread[main,5,main] {istat} stopping daemon, 
active=false, work/age=0/52180 [q/p/s=0/0/0,err:k/u/c=0/0/0,rej:f/d/o=0/0/0]

I did tail -f readlocks/derby.log | grep "{istat}" since the log is deleted 
when the run is successful.

Further, the istat deamon will only run if the following default requirements 
are fulfilled:
 o statistics are created only when the table exceeds the minimum row threshold 
(250 rows)
 o statistics are updated only when the row difference between the two 
estimates exceeds 1000 rows
 o statistics are updated only when the ln difference between the two estimates 
exceeds 1.0.

Since the statistics are index statistics there must be an index on the table, 
and finally a statement involving an index must be prepared (i.e select * from 
t where indexed_col = ?).
                
> readlocks.sql fails with extra locks. Possibly  Automatic Index statistics 
> kicking off during test
> --------------------------------------------------------------------------------------------------
>
>                 Key: DERBY-5599
>                 URL: https://issues.apache.org/jira/browse/DERBY-5599
>             Project: Derby
>          Issue Type: Bug
>            Reporter: Kathey Marsden
>
> I saw this failure for the Feb 1 run at: 
> http://people.apache.org/~myrnavl/derby_test_results/v10_8/linux/testlog/ibm15/1239442-derbyall_diff.txt
> I think it is likely the index statistics update kicking in during the test. 
> I am thinking  should not be disabled for the derbyall store tests as having 
> it kick in can cause upredictable reporting of locks pages used, etc.
> *** Start: readlocks jdk1.5.0 storeall:storemore 2012-02-01 21:11:01 ***
> 11113a11114
> > APP     |UserTran|ROW     |1   |S   |A           |(2,6)     |GRANT|ACTIVE  
> 11122a11124
> > APP     |UserTran|ROW     |1   |S   |A           |(2,6)     |GRANT|ACTIVE  
> 11131a11134
> > APP     |UserTran|ROW     |1   |S   |A           |(2,6)     |GRANT|ACTIVE  
> 11138a11142
> > APP     |UserTran|ROW     |1   |S   |A           |(2,6)     |GRANT|ACTIVE  
> Test Failed.
> *** End:   readlocks jdk1.5.0 storeall:storemore 2012-02-01 21:13:31 ***

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to