Ok, will try new config. Custom is the evaluator that does the harvesting. 
Depending on value of boolean, it looks for new nodes added after the last 
harvest and process only those.

On Thursday, September 4, 2014 3:28:29 AM UTC-6, Michael Hunger wrote:
>
> What is the Custom evaluator?
>
> Am 04.09.2014 um 04:42 schrieb José Cornado <[email protected] 
> <javascript:>>:
>
> I have been busy with other problems 
>
> This is the line that does the traversal description:
>
> graph.traversalDescription().breadthFirst().evaluator(new Custom(true, 
> METHOD_NODE, DynamicRelationshipType.withName(Element.VALUE.name())));
>
> thanks a lot!
> On Monday, September 1, 2014 8:06:05 PM UTC-6, José Cornado wrote:
>>
>> Ok. Between the last time I ran an scenario of 100's of thousands of 
>> nodes and this time, I changed the way data was harvested.
>>
>> Before I was using global graph operations, which made the next step in 
>> our logic inefficient (pushing data through a memory mapped file) because 
>> value nodes could come from anywhere with no common ancestors.
>>
>> So it seems that I will need to have both approaches depending on the 
>> number of nodes? 
>>
>> Would explicitly calling GC every X number of elements processed make 
>> things less expensive on the -Xmx parameter?
>>
>> On Saturday, August 30, 2014 10:25:06 AM UTC-6, José Cornado wrote:
>>>
>>> I do have a log file now.
>>>
>>> The layout(simplified) of the graph follows a class structure
>>>
>>> class A{
>>>
>>> int fld;
>>>
>>> int retInt(){
>>> return 0;
>>> }
>>> int retInt1(){
>>> return 0;
>>> }
>>> }
>>>
>>> (A)-method->(retInt)-pre->(fld)->[(value)....]
>>>   |               |-post->(fld)->[(value)...]
>>>   |-method->(retInt1)-pre->(fld)->[(value)....]
>>>                      |-post->(fld)->[(value)...]
>>>
>>> I use evaluators starting on method nodes. They search for relationships 
>>> where the end node is value. The thousands of nodes are value nodes.
>>>
>>> Thanks!
>>>
>>> On Friday, August 29, 2014 2:43:26 PM UTC-6, Michael Hunger wrote:
>>>>
>>>> Neo4j Embedded runs in the JVM that you use it in and it uses the heap 
>>>> of that VM.
>>>>
>>>> Please share what you do in more detail so that we have any chance to 
>>>> help you out, right no it is too much vague information.
>>>>
>>>> Cheers,
>>>>
>>>> Michael
>>>>
>>>> Am 29.08.2014 um 20:57 schrieb José Cornado <[email protected]>:
>>>>
>>>> Yes, I doubled eclipse's heap and it broke the 100,000k barrier. I 
>>>> increased the size of the backend heap (where all the data goes) to see if 
>>>> it can reach the 200,000.
>>>>
>>>> A more succinct question: does neo(embedded) create a process of its 
>>>> own(thus creating another heap) that can be managed with properties, or 
>>>> simple inherits the heap of the master process?
>>>>
>>>> Thanks!
>>>>
>>>> On Friday, August 29, 2014 11:17:29 AM UTC-6, Michael Hunger wrote:
>>>>>
>>>>> This is already tiny heap for eclipse without neo
>>>>>
>>>>> Sent from mobile device
>>>>>
>>>>> Am 29.08.2014 um 18:56 schrieb José Cornado <[email protected]>:
>>>>>
>>>>> I am using 2.13 and have 2GB of memory. 
>>>>>
>>>>> Eclipse runs with these arguments:
>>>>>
>>>>> -Dosgi.requiredJavaVersion=1.7 -XstartOnFirstThread 
>>>>> -Dorg.eclipse.swt.internal.carbon.smallFonts -XX:MaxPermSize=256m -Xms40m 
>>>>> -Xmx512m -Xdock:icon=../Resources/Eclipse.icns -XstartOnFirstThread 
>>>>> -Dorg.eclipse.swt.internal.carbon.smallFonts
>>>>>
>>>>>
>>>>>
>>>>> On Friday, August 29, 2014 10:31:22 AM UTC-6, José Cornado wrote:
>>>>>>
>>>>>> Hello!
>>>>>>
>>>>>> I have been running into the following problem lately (I swear that 
>>>>>> it didn't happen before):
>>>>>>
>>>>>> I have a tool that lives inside an eclipse feature and the tool uses 
>>>>>> an embedded instance of neo as its store. 
>>>>>>
>>>>>> The data itself looks like a broom (several not only one)
>>>>>>
>>>>>> To retrieve nodes at the end of  relationship X, I use evaluators. 
>>>>>>  When the number of these nodes reaches +80,000 the gc thrashes out.
>>>>>>
>>>>>> The amount of memory left seems sufficient (in excess of 20mb most of 
>>>>>> the time)
>>>>>>
>>>>>> Is there an specific config property that I could tweak to lessen or 
>>>>>> solve this problem?
>>>>>>
>>>>>> Thanks a lot!
>>>>>>
>>>>>
>>>>> -- 
>>>>> You received this message because you are subscribed to the Google 
>>>>> Groups "Neo4j" group.
>>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>>> an email to [email protected].
>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>
>>>>>
>>>> -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "Neo4j" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to [email protected].
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>>
>>>>
> -- 
> You received this message because you are subscribed to the Google Groups 
> "Neo4j" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected] <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to