Fw: question about search indexes and clusters

Hi,


>>
>> Scenario:
>>
>> server1/Jahia1 and server2/Jahia2 are clustered
>> Jahia1: configured for no indexation
>> Jahia2: configured for indexation
>>

and both of them using a same search index file, i.e. :

<init-param>
  <param-name>org.apache.lucene.searchIndexRootDir</param-name>
  <param-value>c:/jahialuceneindex</param-value>
</init-param>


>> When a file(PDF) is uploaded on Jahia1, a JMS message is sent to Jahia2 
>> to index the PDF.
>> Jahia2 indexes the PDF...
>> Question 1: How does it access the PDF?

yes,
all clusters use a same webdav server for storing and retrieving files

>>
>> After Jahia2 finishes the indexation, it sends a JMS message to Jahia1 to 
>> update the index, so
>> that users can search  on Jahia1??
>>

no messages are sent to notify that indexation is done, as it is performed
in a separate Thread in background.
Jahia1 can  see the result once the indexation has been done. In mean time,
search queries are still processed.


Regards,
Khue Nguyen




>> Thanks
>> Gilles
>>
>>
>> --- Khue Nguyen <[email protected]> a écrit :
>>> the search engine uses the clustering capacities of Jahia's cache system 
>>> (
>>> using JMS ) to propagate search indexation between all clustered Jahia
>>> servers.
>>> So when a file upload occurs on a server that actually doesn't index 
>>> files,
>>> the other clusters will ( as long as they are set to do so ).
>>>
>>>
>>> Regards,
>>> Khue Nguyen
>>>
>>>
>>>
>>> ----- Original Message ----- 
>>> From: "Gilles Schlienger" <[email protected]>
>>> To: <[email protected]>
>>> Sent: Friday, January 14, 2005 9:18 PM
>>> Subject: Re: question about search indexes and clusters
>>>
>>>
>>> > Thanks Khue
>>> >
>>> > I understand what you mean,
>>> > but what happens if a user is uploading a file using the server that 
>>> > does
>>> > not index files?
>>> > Or, how do you force users to upload files using only one server?
>>> > If the file is not indexed, you can't search it?
>>> >
>>> > Thanks
>>> > Gilles
>>> >
>>> > --- Khue Nguyen <[email protected]> a écrit :
>>> >> Hi,
>>> >>
>>> >>
>>> >> by default each Jahia instance do index and search in a separate 
>>> >> index
>>> >> file.
>>> >> But you can configure each of them to :
>>> >>
>>> >> 1) only search but no perform indexation
>>> >> 2) to use a common search index
>>> >>
>>> >> These settings are done in the web.xml file of each Jahia instance.
>>> >>
>>> >> <init-param>
>>> >>  <param-name>org.apache.lucene.searchIndexRootDir</param-name>
>>> >>   <!-- to point to a common search index ( but only one Jahia 
>>> >> instance
>>> >> should run indexation -->
>>> >>  <param-value>/WEB-INF/var/search_indexes</param-value>
>>> >> </init-param>
>>> >>
>>> >> <init-param>
>>> >>  <param-name>org.apache.lucene.localIndexing</param-name>
>>> >>  <param-value>1</param-value> <!-- 1: allow local indexing, 0: do not
>>> >> index -->
>>> >> </init-param>
>>> >>
>>> >> we've managed to refactor the search service in near future.
>>> >>
>>> >>
>>> >> Regards,
>>> >> Khue Nguyen
>>> >>
>>> >>
>>> >>
>>> >> ----- Original Message ----- 
>>> >> From: "Gilles Schlienger" <[email protected]>
>>> >> To: <[email protected]>
>>> >> Sent: Friday, January 14, 2005 3:50 PM
>>> >> Subject: question about search indexes and clusters
>>> >>
>>> >>
>>> >> > Hi everyone,
>>> >> >
>>> >> > I'm currently setting up Jahia in a cluster environment with 2 
>>> >> > Jahia
>>> >> > servers.
>>> >> >
>>> >> > I need failover support in my architecture. Thus I need to have 
>>> >> > data
>>> >> > duplicated on both servers.
>>> >> > I'm thinking about implementing a mechanism to notify and 
>>> >> > synchronize
>>> >> > the
>>> >> > other machine and vice
>>> >> > versa, especially for the Jahia files (Big Text, PDFs etc...)
>>> >> >
>>> >> > But what about the search indexes?
>>> >> > I'm a little nervous about synchronizing index files.
>>> >> > I'm thinking about this scenario:
>>> >> > Since indexing a file (PDF) can take some time, if user1 indexes a 
>>> >> > PDF
>>> >> > on
>>> >> > server1 and user2
>>> >> > idnexes a PDF on server2 almost at the same time, then index files
>>> >> > could
>>> >> > be overriden one with the
>>> >> > other during synchronisation.
>>> >> >
>>> >> > I've read the Jahia Clustering HowTo. They talk about only having 
>>> >> > one
>>> >> > Jahia instance for indexing,
>>> >> > ut how do you implement this? (do you configure with the Lucene 
>>> >> > config
>>> >> > files ?)
>>> >> >
>>> >> > Thanks again for any help about clustering
>>> >> > Gilles
>>> >> >
>>> >> >
>>> >> >
>>> >> >
>>> >> >
>>> >> >
>>> >> > Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour 
>>> >> > vos
>>> >> > mails !
>>> >> > Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/
>>> >>
>>> >>
>>> >>
>>> >
>>> >
>>> >
>>> >
>>> >
>>> >
>>> > Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour 
>>> > vos
>>> > mails !
>>> > Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/
>>> >
>>>
>>>
>>>
>>
>>
>>
>>
>>
>>
>> Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour vos 
>> mails !
>> Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/
>>
>
>
>