vivextra at gmail
Sep 18, 2008, 4:13 PM
Post #6 of 8
(294 views)
Permalink
|
Thanks Mike for the insight. I did check the stdout log and found it
was complaining of not having enough disk space. I thought we need
only x2 of the index size. Our index size is 10G (max) and we had 45G
left on that parition - should it still complain of the space?
Some comments/questions on other issues you raised,
We have 2 threads that index the data in two different indexes and
then we merge them into a master index with following call,
masterWriter.addIndexesNoOptimize(indices);
Once the smaller indices have merged into the master index we delete
the smaller indices.
This process runs every 5 minutes. Master Index can grow up to 10G
before we partition it - move it to other directory and start a new
master index.
Every hour we then optimize the master index using,
writer.optimize(optimizeSegment); //where optimizeSegment = 10
Here are my questions,
1) Is this process flawed in terms of performance and efficiency? What
would you recommend?
2) When you say "partial optimize" what do you mean by that?
3) In Lucene 2.3 "segment merging is done in a background thread" -
how does it work, ie, how does it know which segments to merge? What
would cause this background merge exception?
4) Can we turn off "background merge" if I'm running the optimize
every hour in any case? How do we turn it off?
Thanks,
-vivek
On Thu, Sep 18, 2008 at 7:19 AM, Michael McCandless
<lucene[at]mikemccandless.com> wrote:
> Lucene tries to carry forward the root cause exception from the merge,
> into that IOException that optimize throws. But it doesn't always
> succeed in doing so; I'll open a Jira issue and try to figure out why
> this is the case.
>
> All the exception "means" is that the optimize didn't finish -- you
> still have multiple segments in the index. The index should still be
> fine (not corrupt, nothing lost).
>
> What's happening is a BG merge thread is hitting an unhandled
> exception. The JRE will log such unhandled exceptions to System.err
> by default, so, you should scour the app server's logs to find it (it
> should be there).
>
> BTW, that merge that's being attempted is particularly inefficient --
> you are merging an immense segment (the first one) with a bunch of
> tiny ones. A partial optimize could be much better.
>
> Things to try w/o code changes:
>
> * Use a separate tool, eg Luke, to run optimize and see what
> root exception is thrown.
>
> Things to try with code changes:
>
> * Switch to SerialMergeScheduler -- it should still throw the
> exception, but you'll see the full root cause.
>
> * As of 2.4 (coming soon), subclass ConcurrentMergeScheduler and
> override the handleMergeException to do your own logging so you
> that you see the detailed root-cause exception.
>
> * Alternatively, before Lucene 2.4, if you are running in JRE 1.5+
> environment, you can set the default exception handler for
> threads to do your own logging:
>
>
> http://java.sun.com/j2se/1.5.0/docs/api/java/lang/Thread.html#setDefaultUncaughtExceptionHandler(java.lang.Thread.UncaughtExceptionHandler)
>
> Mike
>
> On Sep 17, 2008, at 4:24 PM, vivek sar wrote:
>
>> Hi,
>>
>> We have been running Lucene 2.3 for last few months with our
>> application and all the sudden we have hit the following exception,
>>
>> java.lang.RuntimeException: java.io.IOException: background
>> merge hit exception: _2uxy:c11345949 _2uxz:c150 _2uy0:c150 _2uy1:c150
>> _2uy2:c150 _2uy3:c150 _2uy4:c82 into _2uy5 [optimize]
>>
>> I don't see any other error messages (or stacktrace) around this
>> exception message. This problem doesn't seem to be recoverable and the
>> indexer process is failing even after the reboot of the machine.
>>
>> I've gone through the mailing list over this issue and saw few
>> suggestions,
>>
>> 1) Make sure you've enough disk space (x2 the index size) - our
>> index size is around 5 GB and we have around 50GB space available so
>> this shouldn't be the case
>> 2) Is your machine multi-core - yes, this application is running on
>> Linux box with 8 CPUs, not sure if this is the problem
>>
>> I can't update the code as this is running on the customer site. Here
>> are my questions,
>>
>> a) Is there any workaround to this problem without updating the code
>> base?
>> b) Is there a jira opened on this issue?
>> c) Has this been fixed in the subsequent Lucene releases?
>>
>> Thanks,
>> -vivek
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: java-user-unsubscribe[at]lucene.apache.org
>> For additional commands, e-mail: java-user-help[at]lucene.apache.org
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe[at]lucene.apache.org
> For additional commands, e-mail: java-user-help[at]lucene.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe[at]lucene.apache.org
For additional commands, e-mail: java-user-help[at]lucene.apache.org
|
