Mailing-List: contact java-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: java-user@lucene.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
Message-Id: <36F19444-085E-4001-9C6A-D8C06302B062@mikemccandless.com>
From: Michael McCandless <lucene@mikemccandless.com>
To: java-user@lucene.apache.org
In-Reply-To: <8c4e68610808080223p343f390i1fc3d2a5e42b6a7e@mail.gmail.com>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
Mime-Version: 1.0 (Apple Message framework v926)
Subject: Re: Efficient delete
Date: Fri, 8 Aug 2008 06:27:40 -0400
References: <1f8734680808070546q6d36cbe8m6687b70f7782d6df@mail.gmail.com>
 <8c4e68610808080223p343f390i1fc3d2a5e42b6a7e@mail.gmail.com>


Also, can you do your deletes via IndexWriter (delete by Term) instead  
of opening IndexReader to do the deletes?

Mike

Ian Lea wrote:

> Michael
>
>
> Did you get anywhere with this? 3 secs for one delete is excessive. A
> job of mine ran earlier today and did 2000+ deletes by term on unique
> id in less than 9 seconds.  The index is smaller, at around 5Gb, but I
> don't believe that would explain the difference.  All the deletes were
> done in one batch using lucene 2.3.2.
>
> Is there a lot of other concurrent activity on the index?  On the
> server?  Is it CPU or IO bound?
>
>
> --
> Ian.
>
>
>
> On Thu, Aug 7, 2008 at 1:46 PM, Michael Zehrer <zehrer@zepan.org>  
> wrote:
>> Dear List,
>>
>> I have a rather big index around 20gb. My documents have a unique id
>> that I store in in an untokenized field.
>>
>> Using an IndexReader I delete documents by term using the id. The
>> applications tries to batch as many delete operations as possible for
>> this.
>>
>> The applications runs on a 8Core Machine with 1GB heapsize for the
>> JVM. Each delete takes about 3sec to complete. I have a lot of  
>> Updates
>> during a day so this takes a lot of time and in effect this is too
>> slow.
>>
>> Is there anything I can do to speed things up a little?
>>
>> Best,
>>
>> Michael
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
>> For additional commands, e-mail: java-user-help@lucene.apache.org
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-user-help@lucene.apache.org
>


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org