adequate
adequate
adequate
adequate
JPPF, java, parallel computing, distributed computing, grid computing, parallel, distributed, cluster, grid, cloud, open source, android, .net
JPPF

The open source
grid computing
solution

 Home   About   Features   Download   Documentation   Forums 
November 22, 2017, 11:57:08 AM *
Welcome,
Please login or register.

Login with username, password and session length
Advanced search  
News: New users, please read this message. Thank you!
  Home Help Search Login Register  

Recent Posts

Pages: [1] 2 3 ... 10
1
Forums / JPPF + JPDA
« Last post by jacogeld on Today at 12:49:37 AM »
We have hit a stumbling block in a project that combines JPPF and JPDA.  In essence, we run a  Java program P over and over with different inputs.  As P executes, we monitor it and record its behaviour.  To do this, we use JPDA: a LaunchingConnector starts a separate JVM which our application monitors using the JVM's built-in debugging support.

We want to use JPPF to distribute this task (running P over and over).  It works a treat for our monitoring application, but the VM spawned by JPDA use the usual class loader and tries to load P from the nodes' local disks, instead of from the server.  Is there an existing solution to this use-case?  In other words, is there (a hopefully easy) way to start a VM with a JPPFClassLoader without starting a new node?
2
Forums / Re: Please post a message when you register
« Last post by jacogeld on Today at 12:42:13 AM »
Hello
3
Installation and Configuration / Обращение
« Last post by Harmonic on November 20, 2017, 11:02:59 AM »
Почему существует запрет на открытия в области Математики и Языкознания?
Считаю своим гражданским и общественным долгом обратить Ваше внимание, что экономика - это математика (и все вещи - суть числа, источник, из которого черпает силы (идеи) всё человечество).
https://pravo.media/otkrytye-pisma/pochemu-sushhestvuet-zapret-na-otkrytiya-v-oblasti-matematiki-i-yazykoznaniya/
http://piromi9.wixsite.com/eni-x/research-blog
4
Installation and Configuration / Grails' domains and JPPF
« Last post by sawnd on November 15, 2017, 03:49:17 PM »
Hello,
 
I'm trying to use some Grails domains' methods like Class.findByProperty on a task but i have an exception raised by the nodes:
Code: [Select]
java.lang.ILlegalException: Method on class XXX was used outside of a Grails application.
Is there a special config for the nodes to use those methods?

Thanks
5
Troubleshooting / Re: JPPF Nodes Jobs execution are not in SYNC.
« Last post by lolo on November 08, 2017, 05:55:22 PM »
Hi,

"max.memory.option" is not used any more. It was used in a very old version of JPPF years ago ...
You are correct that passing the -Xmx via the "jppf.jvm.options" property is now the proper way to do it.

Sincerely,
-Laurent
6
Troubleshooting / Re: JPPF Nodes Jobs execution are not in SYNC.
« Last post by kaala on November 08, 2017, 05:33:19 PM »
Hi,

Thanks for quick reply, I will check with most possible way to identify component causing performance issue.

I have one more question, currently I have configured heap size to 4GB using this property max.memory.option =4096 in jppf-node.properties. But when I'm starting the node it is taking only 2GB as max heap size. May I know whether this property is used or not while starting node.

Node build.xml (No change in start scripts)

<!-- to run a single node -->
   <target name="run.node">
      <java fork="yes" classname="org.jppf.node.NodeLauncher" classpathref="run.classpath">
         <jvmarg value="-server" />
         <jvmarg value="-Xmx16m" />
         <jvmarg value="-XX:+UseLargePages" />
         <jvmarg value="-Dlog4j.configuration=log4j-node.properties" />
         <jvmarg value="-Djppf.config=jppf-node.properties" />
      </java>       
   </target>

If I'm passing max memory option as other.jvm.options = -Xms4096m -Xmx4096m in same property file it is taking max heap size as 4GB.

Thanks for your help in advance...

Thanks,
Kaala
7
Troubleshooting / Re: Driver Memory Issue
« Last post by lolo on November 08, 2017, 08:31:27 AM »
Hello,

Thank you very much for providing all this feedback. But mostly I'm glad that you could figure out this issue.
Thanks to you I just learned soemthing about Hibernate, so our time is not wasted, and hopefully your sharing it in the community forums will help other developers as well.

Sincerely,
-Laurent
8
Troubleshooting / Re: Driver Memory Issue
« Last post by arefaydi on November 08, 2017, 08:11:23 AM »
Hello,

Finally resolved, it was about hibernate. I removed cascades from entity and memory usage decreased to 1-2 gb from 10-15 gb and system memory usage normal. We have cascaded entities three level and even we didn't need cascade operation, hibernate keeping references and checking child records if there is any change on them while every operation to parent record. On each level, there are 1000-1500 child records and there is alot of db operation on each task. I think, memory growing was so fast (gigabytes on minutes),  it was also generating memory fragmentation so memory error was occuring even there is memory. I'm really sorry to discuss this issue here and take your time because it was not about jppf.
9
Troubleshooting / Re: JPPF Nodes Jobs execution are not in SYNC.
« Last post by lolo on November 08, 2017, 07:38:51 AM »
Hello,

Without knowing more about your tasks, and assuming all your nodes have a similar configuration and run on equivalent hardware, I would suggest to profile the "slow" nodes while they are executing tasks, using a Java profiler such as JProfiler or YourKit, to find out what is taking time during task execution (i.e. what are the execution hotspots).

If you do not have the possibility to perform profiling, then maybe you could instrument the code of the tasks to print or log timestamps or time measurements at meaningful points of their execution.

The goal is to be able to narrow down the performance bottleneck and figure out which component is causing it, either JPPF, Coherence, your code, Java, etc ...

Sincerely,
-Laurent
10
Troubleshooting / JPPF Nodes Jobs execution are not in SYNC.
« Last post by kaala on November 07, 2017, 08:54:50 PM »
Hi,

I have configured one driver and six nodes for executing task paralley. Divided almost equal six task to six nodes, driver job is just to submit task to nodes and once done sent combined response to server. I have oracle coherence cache to hold data required for each nodes while executing task(each node reading data paralley from cache while executing task).

I'm facing issue nodes task execution are not in sync causing delay in overall JOB, some nodes are taking 30 minutes to finish task and some nodes are taking 5-6 hours(all nodes configuration is same).

Is there I'm missing any configuration or any settings from JPPF configuration side?

Thanks for your help in advance...
Pages: [1] 2 3 ... 10
JPPF Powered by SMF 2.0 RC5 | SMF © 2006–2011, Simple Machines LLC Get JPPF at SourceForge.net. Fast, secure and Free Open Source software downloads