I would recommend you use NIOFSDirectory. We had similar issues and after we
switched to NIOFSDirectory these issues disappeared (dramatically reduced).
Also if you are using filters what has really helped us is to use cached
filters (CachingWrapperFilters).
Hope that helps.
Yannis.
On Oct 6
Hi Everyone,
Recently we have migrated from lucene 2.2 to lucene 2.9.3.
We are having some issues in search. During the load, searchers are getting
hung up. When we took a process stack, we sound that there are around 450
threads are in the blocked state.
Thread t...@885: (state = BLOCKED)
-
o
I do some warmup queries on both machines.
I also checked heap size, it is 1G on 32 bit and 4G on 64 bit machine.
I also have SATA disk on 32 bit and SCSI on 64 bit. So I/O should not be a
problem.
Actually if I'm using MMapDirectory and doing the same query over and over
again I would guess the
Awesome! Thanks a lot Steven! This is exactly what I wanted.
Hi Sirish,
Have you looked at SpanQuery's yet?:
http://lucene.apache.org/java/3_0_2/api/all/org/apache/lucene/search/spans/package-summary.html
See also this Lucid Imagination blog post by Mark Miller:
http://www.lucidimagination.c
Hi Sirish,
Have you looked at SpanQuery's yet?:
http://lucene.apache.org/java/3_0_2/api/all/org/apache/lucene/search/spans/package-summary.html
See also this Lucid Imagination blog post by Mark Miller:
http://www.lucidimagination.com/blog/2009/07/18/the-spanquery/
One common technique, instead
Hmmm... My mistake.
In fact it is not a phrase search, but its a proximity search.
My screen gives four options to the user: -All words, -Exact phrase, -At
least one word, -Within proximity of xx words.
In case of -All words and -At least one word, this is irrelevant an
everything works fine.
I've determined the problem. It's the same end bug as we experienced
with the Cassandra encoding and the term enumeration not being properly
returned. I've outlined the issues in this bug on Lucandra.
http://github.com/tjake/Lucandra/issues/#issue/40
As you can see, the enumeration of the Lucan
Hi Sirish,
I think I understand "within sentence" phrase search - you want the entire
phrase to be within a single sentence. But can you give an example of "non
sentence specific phrase search"? It's not clear to me how useful such
capability would be.
Steve
> -Original Message-
> F
Hello All:
Can any one suggest me the best way to implement both sentence specific and
non sentence specific phrase search? The user is going to have a check box
for phrase search on the screen that says 'within sentence'. If s/he selects
'within sentence', then I should perform sentence specifi
Also note that if you've fired some warmup queries at one machine but not
the other that'll skew your results...
Erick
On Wed, Oct 6, 2010 at 7:22 AM, Sergey wrote:
> Thanks, I'll give it a try.
>
> 06.10.10, 14:47, "Toke Eskildsen" :
>
> > On Wed, 2010-10-06 at 12:22 +0200, Sergey wrote:
> >
> OK. I'll double check the reports.
The issue I'm tackling in this app is that multiple servers are applying
equal-size volumes of updates in a distributed transaction and one means of
node
failure detection is based on detecting a server's failure to prepare a batch
of
updates in a timescal
Thanks, I'll give it a try.
06.10.10, 14:47, "Toke Eskildsen" :
> On Wed, 2010-10-06 at 12:22 +0200, Sergey wrote:
> > When running application on Windows XP 32 bit machine the search time is
> 0.5 second. JVM is IBM Java 5 for 32 bit.
> > But when running the same application on much more pow
On Wed, 2010-10-06 at 12:22 +0200, Sergey wrote:
> When running application on Windows XP 32 bit machine the search time is 0.5
> second. JVM is IBM Java 5 for 32 bit.
> But when running the same application on much more powerfull Windows Server
> 2007 64 bit machine the search time is 3 seconds.
On Wed, Oct 6, 2010 at 12:18 PM, Michael McCandless
wrote:
> There's no quick way to recover from this.
>
> You'd need to derive the segments_N file by opening the files that did
> get written. Conceptually this should be possible (there was a recent
> thread about it), but I don't think anyone h
When running application on Windows XP 32 bit machine the search time is 0.5
second. JVM is IBM Java 5 for 32 bit.
But when running the same application on much more powerfull Windows Server
2007 64 bit machine the search time is 3 seconds. JVM is IBM Java 5 for 64 bit.
It just does not make any
There's no quick way to recover from this.
You'd need to derive the segments_N file by opening the files that did
get written. Conceptually this should be possible (there was a recent
thread about it), but I don't think anyone has actually created the
tool to do so...
Mike
On Wed, Oct 6, 2010 a
Hi,
I was indexing some documents, but my program crashed after several days
of work. If I reopen this index it is empty. I guess the reason is that
auto-commit was not set and I never performed a commit. (Lesson
learned) So probably all documents are marked as "deleted" and
re-opening the i
17 matches
Mail list logo
