Re: blob, single or multi parameters

Mapluz Dev <[email protected]> Fri, 11 Oct 2013 19:24:32 +0200
Newsgroups gmane.comp.web.mnogosearch.general
Organization Mapluz
Message-ID <[email protected]>
thanks a lot for your information

Le 03/10/2013 14:13, Alexander Barkov a écrit :
> Hi,
>
>
> On 10/03/2013 02:53 PM, [email protected] wrote:
>> Hi
>> I have install release mnogosearch 3.3.14 with a mysql database
>> i have create table with thois command :
>>
>> *./indexer -Ecreate -d /usr/local/mnogosearch/etc/indexer.conf*
>>
>> in my indexer .conf i have this :
>>
>> *DBAddr mysql://root:mypassword@localhost/mnogosearchactu/?DBMode=multi*
>>
>> when i try to run search, i have this message :
>> *Inverted word index not found. Probably you forgot to run 'indexer
>> -Eblob'*.
>> so i have run this command :
>>
>> *./indexer -Eblob -d /usr/local/mnogosearch/etc/indexer.conf*
>
> I guess you forgot to fix DBAddr in search.htm to match the one in 
> indexer.conf.
>
>>
>> and all run, but i have questions :
>>
>> 1 - why must i run this commande :*./indexer -Eblob -d 
>> /usr/local/mnogosearch/etc/indexer.conf
>> *      i do not uderstand the*-Eblob*  parameter
>> 2 - in my crontab, is this line correct*  **00 23 * * * 
>> /usr/local/mnogosearch/sbin/indexer -d 
>> /usr/local/mnogosearch/etc/indexer.conf*  to indexing all days at 23h ?*
>>
>
> With DBMode=multi crawling and indexing is done at the same time.
> The advantage is that search index is always up to date with
> what crawler has already downloaded.
>
> With DBMode=blob crawling and indexing are separated in time.
> The advantage of DBMode=blob is that it is much faster at search
> time than DBMode=blob.
> But it needs an extra step "indexer --index"
> (or "indexer -Eblob" - these commands are synonyms)
> to make the index up to date after the crawler has downloaded
> a number of documents with new content
> (i.e. both new documents and old documents that have changed
> since last crawling).
>
>
> The choice between DBMode=multi and DBMode=blob can be done
> depending on the database size and search performance.
>
>
> - If your document collection is rather small and you're
> happy with search performance provided by DBMode=multi,
> then use this command in both indexer.conf and search.htm:
>
> DBAddr mysql://root:mypassword@localhost/mnogosearchactu/?DBMode=multi
>
> The command in crontab is Okey in this case.
>
>
> - If your document collection is rather big, and/or you prefer faster
> search results, then use this DBAddr in both indexer.conf and
> search.htm:
>
> DBAddr mysql://root:mypassword@localhost/mnogosearchactu/?DBMode=blob
>
> In this case, the crontab task should do two things consequently:
>
> # Crawling
> /usr/local/mnogosearch/sbin/indexer -d 
> /usr/local/mnogosearch/etc/indexer.conf
> # Indexing
> /usr/local/mnogosearch/sbin/indexer --index -d 
> /usr/local/mnogosearch/etc/indexer.conf
>
> It's a good idea to put these two commands into a shell script,
> then use it from crontab.
>
>
> Now you can try to change search.htm changing between DBMode=blob
> and DBMode=multi and compare performance.
>
>
>
> If you decide to stay with DBMode=multi, then just copy DBAddr
> from indexer.conf to search.htm.
>
>
> If you decide to switch to DBMode=blob, then it's a good idea
> to start from scratch:
>
> 1. Drop the tables in the current database that were created
> for DBMode=multi
>
> indexer --drop
>
> 2. Edit indexer.conf and search.htm, change to DBMode to blob.
>
> 3. Create tables for DBMode=blob
>
> indexer --create
>
> 4. Crawl your document collection
>
> indexer
>
> 5. Create index
>
> indexer --index
>
> 6. Search
>
>
>
>
>
>> *Thanks a lot for your responses.*
>> *
>>
>>
>>
>>
>>
>> _______________________________________________
>> General mailing list
>> [email protected]
>> http://lists.mnogosearch.org/listinfo/general
>>
>


-- 
VBLC Signature
------------------------------------------------------------------------

Développement Mapluz - MAPLUZ <http://www.mapluz.fr>

Ingénierie Génie Logiciel

*Mobile :*+33 6 79 24 91 50

*Email :*[email protected] <mailto:[email protected]>

_______________________________________________
General mailing list
[email protected]
http://lists.mnogosearch.org/listinfo/general