db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Knut Anders Hatlen (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (DERBY-6215) derby.log grows enormously during StressMultiTest
Date Wed, 08 May 2013 14:55:17 GMT

     [ https://issues.apache.org/jira/browse/DERBY-6215?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Knut Anders Hatlen updated DERBY-6215:

    Attachment: derby-6215-01-a.diff

Other JUnit tests that set a debug flag in the SanityManager seem to use SanityManager.DEBUG_SET()
and SanityManager.DEBUG_CLEAR() instead of the derby.debug.true property.

The attached patch makes ClobReclamationTest use that approach. I monitored the disk space
used while running suites.All with the patch, and now derby.log didn't grow much bigger than
60MB, which is much better than 4GB. I also verified that ClobReclamationTest still prints
DaemonTrace debug trace to derby.log after the change.
> derby.log grows enormously during StressMultiTest
> -------------------------------------------------
>                 Key: DERBY-6215
>                 URL: https://issues.apache.org/jira/browse/DERBY-6215
>             Project: Derby
>          Issue Type: Bug
>          Components: Test
>    Affects Versions:
>            Reporter: Knut Anders Hatlen
>            Assignee: Knut Anders Hatlen
>         Attachments: derby-6215-01-a.diff
> When running suites.All with sane jars, I've seen derby.log grow very big (over 4GB)
during StressMultiTest. I don't see this when StressMultiTest is run separately, only when
it runs as part of suites.All.
> What takes the most space in derby.log is debug output like this
> DEBUG DaemonTrace OUTPUT: Next urgent assignment : org.apache.derby.impl.services.daemon.ServiceRecord@d1294d
> DEBUG DaemonTrace OUTPUT: servicing org.apache.derby.impl.services.daemon.ServiceRecord@d1294d
> DEBUG DaemonTrace OUTPUT: done org.apache.derby.impl.services.daemon.ServiceRecord@d1294d
> DEBUG DaemonTrace OUTPUT: come back from yield
> repeated over and over.
> I suspect there's a test that sets the DaemonTrace debug flag, and that the flag for
some reason doesn't get reset. I've only found one test that sets it: ClobReclamationTest.
It uses a SystemPropertyTestSetup ensure the derby.debug.true property is reset and the engine
rebooted, but I'm not sure rebooting the engine will reset the state of the SanityManager.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message