| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: cluster support |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
ploe
Cruncher Joined: Oct 11, 2005 Post Count: 2 Status: Offline Project Badges: 
|
We have a cluster of old Sun servers at school that is sitting around doing nothing. Was the client designed primarily for the linux desktop user? Or would it be possible to run the client on a cluster?
---------------------------------------- |
||
|
|
FOBioPatel
Cruncher Joined: Apr 10, 2005 Post Count: 21 Status: Offline |
Short Answer: no cluster support to date, as far as I know.
----------------------------------------Solution: Load Boinc on each SUN server independantly. Long Answer: The goal of clustering is to achieve reliability and scalability by interconnecting multiple independent systems. A cluster is a collection of standard, autonomous, machines, configured so that they appear on the network as a single machine. The application and its data can reside anywhere but this is transparent to the application. I would recommend running Boinc on each of the servers within the cluster independantly, only because to date I have not read anything about installing Boinc in any fashion that can utilize clusters. The reason Boinc may not be moving in the direction of utilizing clusters is because problems with SMP (symmetric multiprocessor system) designs are that performance gain as a function of the number of processors is sublinear, particularly with more than 6-8 processors, primarily because of contention for resources such as busses, memory, and devices. Final Answer: I would wait for a WCG Admin to respond, however. I may be wrong about Boinc's cluster support. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Looking at Running BOINC on Solaris ( http://boinc.berkeley.edu/solaris.php ) I see:
The BOINC client software should work on: * Solaris versions 7, 8, 9, or 10 * Any SPARC processor . . . . . . . . . . . . . . These settings should work on machines with up to 20 CPUs. But we do not have any applications configured for Solaris or SPARC. The next platform we will support (not immediately!) is Mac OS X, then we will decide what to do. mycrofth |
||
|
|
ploe
Cruncher Joined: Oct 11, 2005 Post Count: 2 Status: Offline Project Badges:
|
i don't think the machines are running solaris anymore...i'll have to ask.
----------------------------------------from how i understand it, the class doesn't want the cluster to sit idle. i suppose the nodes could run the clients independently but then it wouldn't be acting like a cluster. :/ |
||
|
|
FOBioPatel
Cruncher Joined: Apr 10, 2005 Post Count: 21 Status: Offline |
Running Boinc in a cluster environment may not be as efficient or reliable as it may seem. Let us assume instead of running Boinc on each computer independantly, you manage to deploy a single instance of Boinc and distribute the processing task across your entire cluster of 20 computers. You now have 20 points of failure, and if your cluster begins sending back massive quantities of bad data because of 1 piece of bad RAM in 1 computer in your cluster - your entire cluster will stop receiving projects from worldcommunitygrid.
----------------------------------------Clusters solve the problem of reliability, and scalability. When set up for reliability, there are various means of redelegating work to idle servers in the event of a failure. When set up for scalability you set up Beowulf Cluster to share network resources, disk/file system resources, or processing resources. When dealing with process management in a cluster, you have 3 options: a Batch System, Preemptive Scheduling & Migration, or Fine Grained Control. Let me elaborate on each individually: batch system: job scheduling is left to the programmer (job runner). Generally a queue is used to ssh remote jobs. This is good for highly parallel applications, such as supercomputing or rendering. (To run optimally, you would write a batch script which tells every single computer in your cluster to start Boinc, and keep running it - which is ironically the complicated equivalent of just installing it independantly on each server to begin with.) preemptive scheduling/migration: this is not a part of the Beowulf solution but may be incorporated into the system. Processes can be automatically migrated based on cluster status. This is geared to an environment that is not dedicated to the cluster or one where different types of jobs are run. Two popular packages that provide this capability are Condor (not open source) and Mosix. (This will be extremely weird to set up, since each processor will be saturated with 100% usage anyway, assuming every computer in your cluster is running Boinc.) fine-grained control: programs control their own synchronization and load balancing using MPI and/or PVM libraries and bproc for process dispatch. (This hurts my brain just thinking about it.) Your cluster has a lot to offer the worldcommunitygrid, the sooner you bring it online the better. My recommendation is you load Boinc on each computer separately.  |
|||
|
||||
|
|