db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mayuresh Nirhali (JIRA)" <derby-...@db.apache.org>
Subject [jira] Commented: (DERBY-418) outofmemory error when running large query in autocommit=false mode
Date Tue, 22 Aug 2006 10:44:15 GMT
    [ http://issues.apache.org/jira/browse/DERBY-418?page=comments#action_12429668 ] 
            
Mayuresh Nirhali commented on DERBY-418:
----------------------------------------

I tried the approach suggested by Dan in DERBY-1142, and my version of change looks like as
below,

In GenericLanguageConnectionContext,

	public void addActivation(Activation a) {
		acts.addElement(a);

+              try {
+                  if (acts.size() > 20) {
+                       resetActivations(false);
+                   }
+              } catch(StandardException e) {
+              }
+
		if (SanityManager.DEBUG) {

			if (SanityManager.DEBUG_ON("memoryLeakTrace")) {

				if (acts.size() > 20)
					System.out.println("memoryLeakTrace:GenericLanguageContext:activations " + acts.size());
			}
		}
	}


I thought it will be better to use the method resetActivations as it is meant to do the desired
business for us. ( let me know if there are any negative implications of using this method
here)

This approach did not solve the OOME but differed it for sometime. The effect of this change
is equal to the effect of having singleUseActivation.close in the finalize method of EmbedRS,
as suggested earlier.

Further, and importantly, I found that there still exist some Activation objects with inUse=true.
I guess there is a synchronization issue here due to which EmbedRS.finalize is not called
for all the objects, thus not marking them unused. On several runs, the worst case I found
is that for 100 such select queries about 10% activation objects still stay on the heap. So,
as the finalize method is not executed for many objects, the activation objects are not freed.
Here, it is irrelevent if they are closed directly from within the finalize method or indirectly
by first just marking them unused.

I ran these tests on both jdk1.6 and jdk1.5.

any thoughts on behaviour of finalizer thread in this context.
any inputs on synchronizing the markunused operation ??

> outofmemory error when running large query in autocommit=false mode
> -------------------------------------------------------------------
>
>                 Key: DERBY-418
>                 URL: http://issues.apache.org/jira/browse/DERBY-418
>             Project: Derby
>          Issue Type: Bug
>          Components: Store
>    Affects Versions: 10.1.1.0
>         Environment: I can reproduce this problem on Win2k/ T42 laptop. jdk141. 
>            Reporter: Sunitha Kambhampati
>         Assigned To: Mayuresh Nirhali
>             Fix For: 10.2.1.0
>
>         Attachments: AutoCommitTest.java
>
>
> On the derby-user list,  Chris reported tihs problem with his application and also a
repro for the problem.  I am logging the jira issue so it doesnt get lost in all the mail.
 (http://www.mail-archive.com/derby-user@db.apache.org/msg01258.html)
> ------from chris's post----------
> I'm running a set of ~50000 queries on one table, using inserts and updates, and i want
to be able to roll them back so i turned off autocommit using setAutoCommit(false).  As the
update runs, the memory used by the JVM increases continually until i get the following exception
about 20% of the way through:
> ERROR 40XT0: An internal error was identified by RawStore module.
>    at org.apache.derby.iapi.error.StandardException.newException(StandardException.java)
>    at org.apache.derby.impl.store.raw.xact.Xact.setActiveState(Xact.java)
>    at org.apache.derby.impl.store.raw.xact.Xact.openContainer(Xact.java)
>    at org.apache.derby.impl.store.access.conglomerate.OpenConglomerate.init(OpenConglomerate.java)
>    at org.apache.derby.impl.store.access.heap.Heap.open(Heap.java)
>    at org.apache.derby.impl.store.access.RAMTransaction.openConglomerate(RAMTransaction.java)
>    at org.apache.derby.impl.store.access.RAMTransaction.openConglomerate(RAMTransaction.java)
>    at org.apache.derby.impl.sql.catalog.DataDictionaryImpl.getDescriptorViaIndex(DataDictionaryImpl.java)
>    at org.apache.derby.impl.sql.catalog.DataDictionaryImpl.locateSchemaRow(DataDictionaryImpl.java)
>    at org.apache.derby.impl.sql.catalog.DataDictionaryImpl.getSchemaDescriptor(DataDictionaryImpl.java)
>    at org.apache.derby.impl.sql.compile.QueryTreeNode.getSchemaDescriptor(QueryTreeNode.java)
>    at org.apache.derby.impl.sql.compile.QueryTreeNode.getSchemaDescriptor(QueryTreeNode.java)
>    at org.apache.derby.impl.sql.compile.FromBaseTable.bindTableDescriptor(FromBaseTable.java)
>    at org.apache.derby.impl.sql.compile.FromBaseTable.bindNonVTITables(FromBaseTable.java)
>    at org.apache.derby.impl.sql.compile.FromList.bindTables(FromList.java)
>    at org.apache.derby.impl.sql.compile.SelectNode.bindNonVTITables(SelectNode.java)
>    at org.apache.derby.impl.sql.compile.DMLStatementNode.bindTables(DMLStatementNode.java)
>    at org.apache.derby.impl.sql.compile.DMLStatementNode.bind(DMLStatementNode.java)
>    at org.apache.derby.impl.sql.compile.ReadCursorNode.bind(ReadCursorNode.java)
>    at org.apache.derby.impl.sql.compile.CursorNode.bind(CursorNode.java)
>    at org.apache.derby.impl.sql.GenericStatement.prepMinion(GenericStatement.java)
>    at org.apache.derby.impl.sql.GenericStatement.prepare(GenericStatement.java)
>    at org.apache.derby.impl.sql.conn.GenericLanguageConnectionContext.prepareInternalStatement(GenericLanguageConnectionContext.java)
>    at org.apache.derby.impl.jdbc.EmbedStatement.execute(EmbedStatement.java)
>    at org.apache.derby.impl.jdbc.EmbedStatement.executeQuery(EmbedStatement.java)
>    at vi.hotspot.database.DataInterface._query(DataInterface.java:181)
>    at vi.hotspot.database.DataInterface.query(DataInterface.java:160)
>    at vi.hotspot.database.UpdateManager.updatePartialTable(UpdateManager.java:518)
>    at vi.hotspot.database.UpdateManager.updatePartialTables(UpdateManager.java:619)
>    at vi.hotspot.database.UpdateManager.run(UpdateManager.java:924)
>    at java.lang.Thread.run(Thread.java:534)
> vi.hotspot.exception.ServerTransactionException
>    at vi.hotspot.database.UpdateManager.updatePartialTable(UpdateManager.java:555)
>    at vi.hotspot.database.UpdateManager.updatePartialTables(UpdateManager.java:619)
>    at vi.hotspot.database.UpdateManager.run(UpdateManager.java:924)
>    at java.lang.Thread.run(Thread.java:534)
> Derby is running in standalone mode. 

-- 
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

        

Mime
View raw message