JDBM - ArrayIndexOutOfBoundsException

• Key: DIRSERVER-348
• Type: Bug
• Status: Closed
• Resolution: Won't Fix
• Priority: Major
• Assignee: Alex Karasulu
• Reporter: Simon Temple

Description

I call InitialDirContext every 40 seconds and after about fourty five minutes I get this exception.
 
java.lang.ArrayIndexOutOfBoundsException
 at jdbm/recman/TranslationPage.get(TranslationPage.java:86)
 at jdbm/recman/LogicalRowIdManager.fetch(LogicalRowIdManager.java:135)
 at jdbm/recman/BaseRecordManager.update(BaseRecordManager.java:284)
 at jdbm/recman/CacheRecordManager.updateCacheEntries(CacheRecordManager.java:405)
 at jdbm/recman/CacheRecordManager.commit(CacheRecordManager.java:339)
 at org/apache/ldap/server/partition/impl/btree/jdbm/JdbmIndex.sync(JdbmIndex.java:489)
 at org/apache/ldap/server/partition/impl/btree/jdbm/JdbmDirectoryPartition.sync(JdbmDirectoryPartition.java:248)
 at org/apache/ldap/server/partition/DefaultDirectoryPartitionNexus.sync(DefaultDirectoryPartitionNexus.java:290)
 at org/apache/ldap/server/DefaultDirectoryService.sync(DefaultDirectoryService.java:223)
 at org/apache/ldap/server/jndi/AbstractContextFactory.getInitialContext(AbstractContextFactory.java:82)
 at javax/naming/spi/NamingManager.getInitialContext(NamingManager.java:662)
 at javax/naming/InitialContext.getDefaultInitCtx(InitialContext.java:243)
 at javax/naming/InitialContext.init(InitialContext.java:219)
 at javax/naming/directory/InitialDirContext.<init>(InitialDirContext.java:80)
 

2006-01-24 13:30:44,544 First Call
2006-01-24 14:18:56,444 First Error
 
Restarting normally fixes the problem (for another 40-60 mins).
 
We do not experience these problems on our Windows or UNIX platforms.

Comments

[Simon Temple]

This appears to be a problem with IBMs java optimisation.

The JVM was being invoked with -opt40

    -X -opt<number> optimization level for java program 0(interpret), 10,30, 40

The problem does not occur when optimisation is set to 10 or is not used.

Not using optimisation is now IBMs recommend approach as the JIT will execute code up to 15% faster than that executed with optimisation level 40!

I suggest this issue is closed.

[Emmanuel Lecharny]

waoh !

It's very scarying that an optimisation could be the cause of such a problem... I thought IBM JVM was very solid, but, yes, we can consider that this is possibly the reason why you have such a pb.

What I suggest is that you just keep an eye on this issue, add some more info about the environment (computer, Linux version, JVM exact version), and close the issue.

I'm not sure you can close it, unless we resolve it. If so, just send a mail, we will "resolve" it to "won't fix", then you will be able to close it.

Thanks a lot for having investigating this problem, it was the next on my list after having close another serious issue :)

Just to clarify...

This is a problem on IBM iSeries Midrange Server running OS400 V5R2 (not LINUX)

Our cumulative package PTF level is C4244520

>java -Djava.version=1.4 -version
java version "1.4.2"


I cannot currently close so please mark as "wont fix".

Thanks

[Emmanuel Lecharny]

So this is a won't fix... If it's really a optimisation related pb with IBM JVM 1.4 on OS/400 system, I'm afraid we won't be able to fix IBM JVM ;)

[Simon Temple]

No action required.

[Emmanuel Lecharny]

moved it and linked it to the correct issue