spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Wush Wu <w...@bridgewell.com>
Subject Re: Run spark on mesos remotely
Date Fri, 28 Mar 2014 01:32:51 GMT
Dear Rustagi,

Thanks for you response.

As far as I know, the DAG scheduler should be a part of spark. Therefore,
should I do something not mentioned in
http://spark.incubator.apache.org/docs/0.8.1/running-on-mesos.html to
launch the DAG scheduler?

By the way, I also notice that the user of the submission becomes the
account on my machine. I doubt that there is a permission issue, so the
executor cannot copy the file from spark.execute.uri. After changing the
account to the same username on the mesos cluster, the executor
successfully copy the file from  spark.execute.uri, but it reports the
following error message:

log4j:WARN No appenders could be found for logger
(org.apache.spark.executor.MesosExecutorBackend).
log4j:WARN Please initialize the log4j system properly.
log4j:WARN See http://logging.apache.org/log4j/1.2/faq.html#noconfig for
more info.
org.apache.spark.SparkException: Error sending message to
BlockManagerMaster [message =
RegisterBlockManager(BlockManagerId(201403250945-3657629962-5050-10180-16,
pc104, 42356, 0),339585269,Actor[akka://spark/user/BlockManagerActor1])]
 at
org.apache.spark.storage.BlockManagerMaster.askDriverWithReply(BlockManagerMaster.scala:174)
at
org.apache.spark.storage.BlockManagerMaster.tell(BlockManagerMaster.scala:139)
 at
org.apache.spark.storage.BlockManagerMaster.registerBlockManager(BlockManagerMaster.scala:57)
at org.apache.spark.storage.BlockManager.initialize(BlockManager.scala:127)
 at org.apache.spark.storage.BlockManager.<init>(BlockManager.scala:105)
at org.apache.spark.storage.BlockManager.<init>(BlockManager.scala:119)
 at
org.apache.spark.SparkEnv$.createFromSystemProperties(SparkEnv.scala:171)
at org.apache.spark.executor.Executor.<init>(Executor.scala:111)
 at
org.apache.spark.executor.MesosExecutorBackend.registered(MesosExecutorBackend.scala:58)
Caused by: java.util.concurrent.TimeoutException: Futures timed out after
[10000] milliseconds
 at akka.dispatch.DefaultPromise.ready(Future.scala:870)
at akka.dispatch.DefaultPromise.result(Future.scala:874)
 at akka.dispatch.Await$.result(Future.scala:74)
at
org.apache.spark.storage.BlockManagerMaster.askDriverWithReply(BlockManagerMaster.scala:160)
 ... 8 more
Exception in thread "Thread-0"

Is there any suggestion to handle the error above?

Thanks,
Wush


2014-03-28 4:09 GMT+08:00 Mayur Rustagi <mayur.rustagi@gmail.com>:

> Yes but you have to maintain connection of that machine to the master
> cluster as the driver with DAG scheduler runs on that machine.
> Regards
> Mayur
>
> Mayur Rustagi
> Ph: +1 (760) 203 3257
> http://www.sigmoidanalytics.com
> @mayur_rustagi <https://twitter.com/mayur_rustagi>
>
>
>
> On Thu, Mar 27, 2014 at 4:09 AM, Wush Wu <wush@bridgewell.com> wrote:
>
>> Dear all,
>>
>> We have a spark 0.8.1 cluster on mesos 0.15. It works if I submit the job
>> from the master of mesos. That is to say, I spawn the spark shell or launch
>> the scala application on the master of mesos.
>>
>> However, when I submit the job from another machine, the job will lost.
>> The logs shows that the mesos does not copy the
>> spark-0.8.1-incubating.tar.gz to the temporal working directory, so the job
>> lost immediately. Is it possible to submit the job from the machine not
>> belong to mesos cluster?
>>
>> Thanks!
>>
>> Wush
>>
>
>

Mime
View raw message