jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dennis van der Laan <d.g.van.der.l...@rug.nl>
Subject Re: Nodes not returned in query when indexing?
Date Tue, 20 Oct 2009 12:05:14 GMT
Hi Marcel (or anyone else on the list),

Can anybody confirm this is a bug? Has it been filed yet and if so, what
is the JIRA issue id?

Thanks,
Dennis

Dennis van der Laan wrote:
> Marcel Reutegger wrote:
>   
>> Hi,
>>
>> On Tue, Sep 29, 2009 at 10:38, Dennis van der Laan
>> <d.g.van.der.laan@rug.nl> wrote:
>>   
>>     
>>> Hi,
>>>
>>> I believe file nodes get 'hidden' somehow when their contents get indexed.
>>> I create a file in a repository, and set a custom property 'myns:id' on
>>> the jcr:content node. Directly after creating the file, I perform an
>>> XPath query retrieving the file node based on the id-property.
>>> Sometimes, especially with (larger) PDF-files, the node can not be found
>>> by this id-property.
>>>
>>> I changed my code, by executing the query over and over again and
>>> waiting for 2 seconds between each try. Eventually the node is returned,
>>> mostly after 4-6 seconds (so 2 to 3 tries). In the log file, when the
>>> query does not return the node, I see the message "MultiIndex: updating
>>> index with 1 nodes from indexing queue. (MultiIndex.java, line 1210)",
>>> letting me believe the node is somehow hidden or locked when it is being
>>> indexed.
>>>
>>> I have seen this behavior when using a direct repository and when using
>>> RMI. It happens when using a local filesystem and when using a database
>>> store, with Java Derby and with Oracle 10.
>>>
>>> My questions:
>>> - Is this 'normal' behavior or is it a bug?
>>>     
>>>       
>> this sounds like a bug. what version of jackrabbit are you using?
>>   
>>     
> I have the same behavior using Jackrabbit 1.5.6 and 1.6.0.
> I discovered setting the extractorPoolSize to 0 in my repository
> configuration 'fixes' the problem, but this has a significant impact on
> the upload performance.
>
> best regards,
> Dennis
>   
>> regards
>>  marcel
>>
>>   
>>     
>>> - How can I prevent this? Would it e.g. help if I exclude the
>>> id-property from being indexed, or will the complete jcr:content node be
>>> hidden when the jcr:data property is being indexed?
>>>
>>> Thanks in advance,
>>> Dennis van der Laa
-- 
Dennis van der Laan


Mime
  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message