jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From François Cassistat...@maya-systems.com>
Subject Re: getRootNode() takes 27 seconds...
Date Sat, 27 Feb 2010 19:32:28 GMT
Stefan,

it seems that you are right.

When using TransientRepository instead of jcr2spi, getRootNode() or getNode("/") is returned
in 13ms.

Somehow, it's weird that I have this problem only with the root node. I was having this problem
without doing anything between getSession and getRootNode.


François



Le 2010-02-27 à 10:17 AM, Stefan Guggisberg a écrit :

> 2010/2/26 François Cassistat <f@maya-systems.com>:
>> Thanks Rakesh for your reply,
>> 
>> actually root have only one node, below 2 or 3 levels of that of that there is one
node with 1500 nodes, this is the maximum. All other nodes have between 0 and 10 nodes.
>> 
>> 1500 may be a lot
> 
> no, it's not ;), and 20k child nodes shouldn't be a problem either
> with your configuration (embedded derby).
> 
> cheers
> stefan
> 
>> and I may implements a tree strategy to decrease the number of child of this node.
But does that sub-sub-sub-sub-node make the getRootNode() going that slow?
> 
> if you're using a remote client which does deep bulk fetches,
> Session.getRootNode() might not just get the root node from the
> server, but a subtree in order to minimize the number of subsequent
> server-roundtrips when traversing the hierarchy.
> 
> cheers
> stefan
> 
>> 
>> 
>> F
>> 
>> 
>> Le 2010-02-26 à 1:16 PM, Rakesh Vidyadharan a écrit :
>> 
>>> 
>>> On 26 Feb 2010, at 11:32, François Cassistat wrote:
>>> 
>>>> I have inserted ~25K nodes with ~1500 files (about 1.6G of files) and now
Jackrabbit takes 27 seconds to get the root node, is it normal? Maybe it is because am I using
the standalone server (2.0)? Also I am making tests since jcr2spi.
>>>> 
>>>> I'm a newbie with JackRabbit configuration files, so I did not make much
modifications to repository.xml
>>> 
>>> In my experience the biggest issue is the number of child nodes under a parent
(and not repository configuration).  The general recommendation is to try and keep child nodes
low (1000 does not seem too bad in my experience), but not 25K under a single parent.
>>> 
>>> Rakesh
>> 
>> 


Mime
View raw message