Re: Index performance

2007-04-19 Thread Doron Cohen
"Tony Qian" wrote on 19/04/2007: > I found the problem which slowed down indexing. It is our NFS file system. If only the index is maintained on NFS (say input is on local disk or DB or such) it may help to index to local disk and only occasionally (once a day?) copy/update to an index maintained

Re: Index performance

2007-04-19 Thread Michael McCandless
"Tony Qian" <[EMAIL PROTECTED]> wrote: > I found the problem which slowed down indexing. It is our NFS file > system. NFS performance is generally slower than local filesystem, though there may be ways to tune it (I'm not sure). I have heard but not personally verified that mounting NFS read-onl

Re: Index performance

2007-04-19 Thread Tony Qian
Doron and Erick, I found the problem which slowed down indexing. It is our NFS file system. Thanks for help. Tony From: "Tony Qian" <[EMAIL PROTECTED]> Reply-To: java-user@lucene.apache.org To: java-user@lucene.apache.org Subject: Re: Index performance Date: Mon, 16 Apr 200

Re: Index performance

2007-04-16 Thread Tony Qian
Doron, I'll try that and let you know the result. thanks for suggestions. Tony From: Doron Cohen <[EMAIL PROTECTED]> Reply-To: java-user@lucene.apache.org To: java-user@lucene.apache.org Subject: Re: Index performance Date: Thu, 12 Apr 2007 13:40:07 -0700 To cover all possible n

Re: Index performance

2007-04-16 Thread Tony Qian
Erick, Sorry for late reply. I was stuck with other project. content object is plain Java object. It has all fields set. Thanks, Tony From: "Erick Erickson" <[EMAIL PROTECTED]> Reply-To: java-user@lucene.apache.org To: java-user@lucene.apache.org Subject: Re: Index perfo

Re: Index performance

2007-04-12 Thread Doron Cohen
To cover all possible non-indexing overhead, better measure with something like this: static long indexContents(IndexWriter writer, List storyContentList) throws IOException { long res = 0; if (storyContentList != null && storyContentList.size() != 0) { try {

Re: Index performance

2007-04-12 Thread Erick Erickson
could that be where the time is spent? Erick On 4/12/07, Tony Qian <[EMAIL PROTECTED]> wrote: Otis, I timed just for indexing. thanks, Tony >From: Otis Gospodnetic <[EMAIL PROTECTED]> >Reply-To: [EMAIL PROTECTED] >To: [EMAIL PROTECTED] >Subject: Re: Index performance >D

Re: Index performance

2007-04-12 Thread Doron Cohen
> I tried to index it. It took from 7-10 seconds to index about 90 documents. That would be around 10 documents per second - way too slow. A Lucene's perf test adding 12,000 docs sized similar to your sample doc (1400 characters) on a not so strong machine shows much faster pace - 146 docs per sec

Re: Index performance

2007-04-12 Thread Tony Qian
Otis, I timed just for indexing. thanks, Tony From: Otis Gospodnetic <[EMAIL PROTECTED]> Reply-To: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Re: Index performance Date: Thu, 12 Apr 2007 09:31:49 -0700 (PDT) Hi Tony, Your code looks fine to me. I'm not sure what you timed

Re: Index performance

2007-04-12 Thread Tony Qian
Eric, Thanks for the information. The id is generated by database and it is unique. So I only need to index it and don't need to store it, right Tony From: "Erick Erickson" <[EMAIL PROTECTED]> Reply-To: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: Re: Index perf

Re: Index performance

2007-04-12 Thread Erick Erickson
Another question is if I can delete document based on storyIndentity field ( using IndexReader.deleteDocuments(term)). Since storyIdentity field is not indexed, is there any performance issue or I should index it too (and store it)? As to your very last question, No, there'll be no performance

Re: Index performance

2007-04-12 Thread Otis Gospodnetic
Hi Tony, Your code looks fine to me. I'm not sure what you timed - the whole app run, just indexing, indexing + optimizing... If you times indexing + optimizing, leave optimization out of the timer. How long do you think this should take? Try setting maxBufferedDocs to 90. Otis . . . . .