jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Singer <jackrab...@it-specialist.at>
Subject Re: [jira] Commented: (JCR-623) Clustering
Date Fri, 10 Nov 2006 08:22:08 GMT
Hi Dominique,

Dominique Pfister wrote:
> On 11/9/06, Pankaj Gupta (JIRA) <jira@apache.org> wrote:
>>     [
>> http://issues.apache.org/jira/browse/JCR-623?page=comments#action_12448517
>> ]
>>
>> Pankaj Gupta commented on JCR-623:
>> ----------------------------------
>>
>> Thanks for geting started on this issue. Its a high priority one for
>> us and perhaps many other Jackrabbit users.
>>
>> Couple of questions:
>> 1) Is the search index in other clusters also synced up if each
>> cluster has the index in its local disk? Or is it required to have the
>> search index  in a common file system?
> 
> No, the index is synced up automatically, as a result of the events -
> passed inside the journal record - being delivered. In other words,
> the search manager, responsible for updating the index, doesn't see a
> difference between updates made by its own node and changes made by a
> foreign node.

What happens if a clusternode fails and is restored after a while? When
the active nodes made changes? Is the index resynced on repository startup?

>> 2) Are the other clusters updated immediately? For example, if one
>> cluster adds a new node, is it immediately visible to other sessions
>> in other clusters, as is mandated by the JCR spec?
> 
> Not actively. Other nodes in the cluster see the change inside a given
> delay (5 seconds is the default) unless they itself try to perform
> some changes, which causes an immediate sync.
> 
> Kind regards
> Dominique

-- 
kind regards

Michael

Mime
View raw message