NullPointerException in <DaemonThread name="Cleaner-1"/> (BDB Java Edition 5.0.73)

2659311

Has anyone come accross the following error:

SEVERE: <DaemonThread name="Cleaner-1"/> caught exception, java.lang.NullPointerException Continuing
java.lang.NullPointerException
        at com.sleepycat.je.cleaner.FileProcessor.processFile(FileProcessor.java:669)
        at com.sleepycat.je.cleaner.FileProcessor.doClean(FileProcessor.java:289)
        at com.sleepycat.je.cleaner.FileProcessor.onWakeup(FileProcessor.java:148)
        at com.sleepycat.je.utilint.DaemonThread.run(DaemonThread.java:163)
        at java.lang.Thread.run(Thread.java:679)

I am using BDB Java Edition 5.0.73.

Greybird-Oracle

No, we haven't seen this.  And when I look at the 5.0.73 source code, I don't see how this could happen.

Did you build the je.jar from sources?  If so could you please confirm that in your sources, FileProcessor.java line 669 is the following?:

                    lookAheadCache.add

                        (Long.valueOf(DbLsn.getFileOffset(logLsn)),

                         new LNInfo(targetLN, dbId, key));

The lookAheadCache variable cannot be null at this point, which is why I say I don't understand how it could happen.

Thanks,

--mark

2659311

No, we did not build it from sources. We used the binary distribution. We have seen this error just a few times. Is it something we should worry about?

Here is the code you requested:

                    lookAheadCache.add

                        (Long.valueOf(DbLsn.getFileOffset(logLsn)),

                         new LNInfo(targetLN, dbId, key));

Greybird-Oracle

As I say, we've never seen it before.  Since it "should never happen" if the JVM is working correctly, I'm not sure what we should do about it.

What JVM and version are you using?

The impact is that the cleaner won't finish cleaning that particular file, but will try again later.  So as long as the cleaner continues, and disk space is reclaimed as normal, then you may be able to get away with ignoring it.  I suppose it depends on how often it happens.

--mark

2659311

We are using Azul Zing 5.7.1. It has happened a few times (2 or 3). We have a heavily loaded prod environment that handles 400MM - 500MM messages daily. We will contact Azul for support...

2659311

BTW, what is the default value of lookAheadCacheSize?

Greybird-Oracle

I see.  As you probably know, we haven't qualified JE against the Azul JVM.

Greybird-Oracle

The FileProcessor.process file method looks something like this (see src/com/sleepycat/je/cleaner/FileProcessor.java for the full method):

    private boolean processFile(...boolean calcUtilizationOnly...) {

...

        final LookAheadCache lookAheadCache = calcUtilizationOnly ?

            null : new LookAheadCache(lookAheadCacheSize);

...

            while (...) {

...

                if (calcUtilizationOnly) {

                    continue;

                }

...

                    lookAheadCache.add

                        (Long.valueOf(DbLsn.getFileOffset(logLsn)),

                         new LNInfo(targetLN, dbId, key));

The NullPointerException occurs when lookAheadCache is referenced above, so it must be null.  But it cannot be null at that point, if the JVM is working correctly, because of the 'continue' statement above it.

--mark