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] Commented: (DERBY-2176) OutOfMemoryError in Network Server after running DOTS-like test for weeks
Date Thu, 14 Dec 2006 15:47:21 GMT
    [ http://issues.apache.org/jira/browse/DERBY-2176?page=comments#action_12458519 ] 
Knut Anders Hatlen commented on DERBY-2176:

I'm just guessing here, but the histogram of objects seems similar to what I would expect
from DERBY-912, which is a tiny memory leak that did not get fixed in (Should be
fixed in though.) That bug caused a vector to grow by one for each sort operation
performed on a connection, and the space would not be freed until the connection was closed.
So if the DOTS test reuses connections and executes a query which performs a sort every now
and then, it will very slowly run out of memory. The symptoms fit because the object histogram
shows that most of the heap space is used for Object[] (because the vector has grown) and
that there are instances of SortResultSet.

> OutOfMemoryError in Network Server after running DOTS-like test for weeks
> -------------------------------------------------------------------------
>                 Key: DERBY-2176
>                 URL: http://issues.apache.org/jira/browse/DERBY-2176
>             Project: Derby
>          Issue Type: Bug
>          Components: Network Server
>    Affects Versions:
>         Environment: Solaris 10 x86_64 MU2, remote client-server, 2 x 2.4 GHz AMD Opteron,
2048 MB RAM, Sun HotSpot 1.5.0_07 server JVM
>            Reporter: John H. Embretsen
>         Attachments: derby.log, jmap-histo_server_17Nov1020CET.txt, OU_jstat-gc-t.png,
> Running a modified version of the Database Opensource Test Suite (DOTS) (v1.1.1) test
case ATCJ2 (auction "simulation"), available from http://sourceforge.net/project/showfiles.php?group_id=3382,
resulted in an "OutOfMemoryError: Java heap space" after 57 days, 13 hours and 21 minutes.
> The test run has previously been described on derby-dev: http://www.nabble.com/Long-running-testing-of-
> The modifications to the original test case were mainly to avoid the type of errors described
in http://www.nabble.com/OutOfMemoryErrors-when-testing-Derby-with-DOTS-t1010027.html. Indeed,
this is a different, yet similar error.
> Heap size (both client and server JVMs) was:
>   -Xms128m
>   -Xmx128m
> The test was run against the release candidate, but since this version does
not exist in Jira, and the delta between and the released is minimal,
is marked as the "affects version".

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message