Re: Running Giraph on YARN (0.23)

[Claudio Martella]

Are you running with an external ZK, or are you expecting giraph to start
one from you on the master? If the second one is the case (as it seams to
me from the logs), then it won't work. Giraph on Yarn requires you to start
ZK externally, and pass the address to giraph when you start the job.


On Thu, Nov 21, 2013 at 11:49 AM, Gianmarco De Francisci Morales <
g...@apache.org> wrote:

> Hi,
>
> I am trying to run Giraph on a Hadoop 0.23 installation, but I am running
> in a communication problem. From the log it seems that the workers are not
> able to connect to the zookeeper instance on the master.
> I am using the option -DuseNetty=true
>
> Any help is appreciated.
> Here the error log:
>
>
>
> 
>
>
> 2013-11-20 15:22:22,000 INFO [main] org.apache.giraph.graph.GraphTaskManager:
> setup: classpath @ /user/bordino/.staging/job_1382563758657_470916/job.jar for
> job Giraph: org.apache.giraph.examples.RandomWalkWithRestartComputation
> 2013-11-20 15:22:22,003 WARN [main] org.apache.hadoop.conf.Configuration:
> mapred.task.partition is deprecated. Instead, use mapreduce.task.partition
> 2013-11-20 15:22:22,003 WARN [main] 
> org.apache.hadoop.conf.Configuration:mapred.job.id is deprecated. Instead, 
> use mapreduce.job.id
> 2013-11-20 15:22:22,004 WARN [main] org.apache.hadoop.conf.Configuration:
> job.local.dir is deprecated. Instead, use mapreduce.job.local.dir
> 2013-11-20 15:22:22,007 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> createCandidateStamp: Made the directory
> _bsp/_defaultZkManagerDir/job_1382563758657_470916
> 2013-11-20 15:22:22,008 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> createCandidateStamp: Made the directory
> _bsp/_defaultZkManagerDir/job_1382563758657_470916/_zkServer
> 2013-11-20 15:22:22,011 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> createCandidateStamp: Creating my filestamp
> _bsp/_defaultZkManagerDir/job_1382563758657_470916/_task/gsta60065.tan.ygrid.yahoo.com
> 56
> 2013-11-20 15:22:22,027 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> getZooKeeperServerList: For task 56, got file 'null' (polling period is 3000)
> 2013-11-20 15:22:25,030 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> getZooKeeperServerList: For task 56, got file
> 'zkServerList_gsta60065.tan.ygrid.yahoo.com 56 '
> (polling period is 3000)
> 2013-11-20 15:22:25,030 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> getZooKeeperServerList: Found
> [gsta60065.tan.ygrid.yahoo.com, 56] 2 hosts in
> filename
> 'zkServerList_gsta60065.tan.ygrid.yahoo.com 56 '
> 2013-11-20 15:22:25,031 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> onlineZooKeeperServers: Trying to delete old directory
> /grid/0/tmp/yarn-local/usercache/bordino/appcache/application_1382563758657_470916/work/_bspZooKeeper
> 2013-11-20 15:22:25,038 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> generateZooKeeperConfigFile: Creating file
> /grid/0/tmp/yarn-local/usercache/bordino/appcache/application_1382563758657_470916/work/_bspZooKeeper/zoo.cfg
> in
> /grid/0/tmp/yarn-local/usercache/bordino/appcache/application_1382563758657_470916/work/_bspZooKeeper
> with base port 22181
> 2013-11-20 15:22:25,038 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> generateZooKeeperConfigFile: Make directory of _bspZooKeeper = true
> 2013-11-20 15:22:25,039 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> generateZooKeeperConfigFile: Delete of zoo.cfg = false
> 2013-11-20 15:22:25,041 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> onlineZooKeeperServers: Attempting to start ZooKeeper server with command
> [/home/Releases/gridjdk-1.7.0_17.1303042057-20130415-000/share/gridjdk-1.7.0_17/jre/bin/java,
> -Xmx128M, -XX:ParallelGCThreads=4, -XX:+UseConcMarkSweepGC,
> -XX:CMSInitiatingOccupancyFraction=70, -XX:MaxGCPauseMillis=100, -cp,
> /user/bordino/.staging/job_1382563758657_470916/job.jar,
> org.apache.zookeeper.server.quorum.QuorumPeerMain,
> /grid/0/tmp/yarn-local/usercache/bordino/appcache/application_1382563758657_470916/work/_bspZooKeeper/zoo.cfg]
> in directory
> /grid/0/tmp/yarn-local/usercache/bordino/appcache/application_1382563758657_470916/work/_bspZooKeeper
> 2013-11-20 15:22:25,044 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> onlineZooKeeperServers: Shutdown hook added.
> 2013-11-20 15:22:25,044 INFO [main] org.apache.giraph.zk.ZooKeeperManager:
> onlineZooKeeperServers: Connect attempt 0 of 10 max trying to connect 
> togsta60065.tan.ygrid.yahoo.com:22181 with poll
> msecs = 3000
> 2013-11-20 15:22:25,048 WARN [main] org.apache.giraph.zk.ZooKeeperManager:
> onlineZooKeeperServers: Got ConnectException
> java.net.ConnectException: Connection refused
> at java.net.PlainSocketImpl.socketConnect(Native Method)
> at
> java.net.AbstractPlainSocketImpl.doConnect(AbstractPlainSocketImpl.java:339)
> at
> java.net

Re: Running Giraph on YARN

[Milinda Pathirage]

Hi Devs,

I was able to get Giraph running on Yarn by creating Yarn specific
configuration programmatically. I think that it's better if we can
have some shell scripts specific to Yarn (or may be modifications to
existing shell scripts), so that we can easily deploy Giraph jobs on
Yarn clusters. Please let me know if anyone is working on this. If no
one is working on that, I would like to work on that.

Thanks
Milinda

On Tue, Aug 13, 2013 at 2:33 PM, Milinda Pathirage
 wrote:
> Hi,
>
> I'm trying to get Giraph running on YARN based on TestYarnJob test
> case. But having issues with moving required jars to YARN environment.
> I'm using single node YARN setup. I can see the job in YARN, but with
> following error.
>
> "Error: Could not find or load main class
> org.apache.giraph.yarn.GiraphApplicationMaster"
>
> I found that FileSystem.get(giraphConf) returns LocalFS inside
> resource copy methods by debugging. Can someone please point me to a
> doc or some writeup which describes how to properly configure
> GiraphYarnClient?
>
> Thanks
> Milinda
>
> --
> Milinda Pathirage
>
> twitter: milindalakmal
> skype: milinda.pathirage
> blog: http://milinda.pathirage.org



-- 
Milinda Pathirage

twitter: milindalakmal
skype: milinda.pathirage
blog: http://milinda.pathirage.org