Hi Supun,

do you find any workers registered as executors registered in the app's UI
(click on the app ID in the spark UI)?

rgds

On Mon, Mar 30, 2015 at 2:27 PM, Supun Sethunga <[email protected]> wrote:

> Hi Niranda,
>
> After the ofline discussion we had, I tried increasing the Spark cores to
> 10, and limiting CarbonAnalytics App's cores to 5, Leaving rest of the 5
> cores to be used by any external app. Now both the apps (CarbonAnalytics
> app and external app) are shown as "Running" in the Spak-Master UI, But I
> keep getting the error at the client side "TaskSchedulerImpl: Initial job
> has not accepted any resources; check your cluster UI to ensure that
> workers are registered and have sufficient memory" .
>
> Further investigating, found out that the worker node listed in the
> Spak-Master UI points to *localhost:8091 *where  in the worker's console
> it says "Analytics worker started: [10.100.5.116:4503:*8092*]" . (I can
> access the worker nodes UI's in both ports *localhost:8091 *and
> *localhost:8092*). It seems there are two worker nodes get initiated, and
> the jobs are passed to the wrong worker. Searching for the cause of the
> error I mentioned earlier (*TaskSchedulerImpl: Initial job has not
> accepted any resources; check your cluster UI to ensure that workers are
> registered and have sufficient memory"*) seems also because of the
> improper connectivity between the master and the worker.
>
> Could you please take look whats happening?
>
> Thanks,
> Supun
>
> On Tue, Mar 10, 2015 at 12:22 PM, Niranda Perera <[email protected]> wrote:
>
>> Okay Anjana, noted!
>>
>> On Tue, Mar 10, 2015 at 12:10 PM, Anjana Fernando <[email protected]>
>> wrote:
>>
>>> Hi Niranda,
>>>
>>> When we get the clustering to properly to work, which would be before
>>> Friday's release. It should be straightforward for ML to point to our
>>> master server and submit jobs. So lets wait till we finalize our
>>> implementation, and then ML team can work with us to get their use case
>>> working.
>>>
>>> Cheers,
>>> Anjana.
>>>
>>> On Tue, Mar 10, 2015 at 10:09 AM, Niranda Perera <[email protected]>
>>> wrote:
>>>
>>>> @supun I suggest you use an external spark cluster for the moment. as
>>>> of the current BAM, we still have not finalized external job submission
>>>> strategy in a BAM cluster. I will discuss this with Anjana and get back to
>>>> you!
>>>>
>>>> On Tue, Mar 10, 2015 at 10:02 AM, Supun Sethunga <[email protected]>
>>>> wrote:
>>>>
>>>>> [looping Nirmal]
>>>>>
>>>>> On Tue, Mar 10, 2015 at 10:01 AM, Supun Sethunga <[email protected]>
>>>>> wrote:
>>>>>
>>>>>> [looping Nirmal]
>>>>>>
>>>>>> Hi Niranda,
>>>>>>
>>>>>> Thanks for the clarification.
>>>>>>
>>>>>> @Nirmal : Seems we have to wait till the BAM M2 release for ML-BAM
>>>>>> Integration :) . But we still can enable to use an external spark cluster
>>>>>> for ML.
>>>>>>
>>>>>> Regards,
>>>>>> Supun
>>>>>>
>>>>>> On Tue, Mar 10, 2015 at 9:49 AM, Niranda Perera <[email protected]>
>>>>>> wrote:
>>>>>>
>>>>>>> cc+ Anjana
>>>>>>>
>>>>>>> Hi Supun,
>>>>>>>
>>>>>>> As per the M1 release, we have only the standalone 'local' mode
>>>>>>> spark instance. as you correctly said, we have set the master to 
>>>>>>> 'local' in
>>>>>>> the BAM . please refer init() method here [1]. In the init() method, we 
>>>>>>> are
>>>>>>> starting a spark context.
>>>>>>>
>>>>>>> we are still working on spark clustering. the plan is to release it
>>>>>>> in the M2.
>>>>>>>
>>>>>>> [1]
>>>>>>> https://github.com/wso2/carbon-analytics/blob/master/components/analytics/analytics-processors/org.wso2.carbon.analytics.spark.core/src/main/java/org/wso2/carbon/analytics/spark/core/AnalyticsExecutionService.java
>>>>>>>
>>>>>>> On Mon, Mar 9, 2015 at 5:48 PM, Supun Sethunga <[email protected]>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I tried starting up a BAM server, and submitting a spark-job to the
>>>>>>>> Spark which is built in BAM, from an external java client. In doing 
>>>>>>>> this, I
>>>>>>>> was not able to create a Spark Context using BAM's spark instance as
>>>>>>>> master. (AFAIK, the way to do this is to create a Spark context inside 
>>>>>>>> the
>>>>>>>> client, and point BAM's Spark instance as the master, and then submit 
>>>>>>>> any
>>>>>>>> jobs to the created context.)
>>>>>>>>
>>>>>>>> Possible reason might be, in BAM, Spark instance (i.e.
>>>>>>>> SparkContext) is started locally with one worker thread.(setting 
>>>>>>>> master as
>>>>>>>> 'local'). Hence the spark instance cannot be accessed by any external
>>>>>>>> client. (Refer attached image "BAM-spark.png")
>>>>>>>> AFAIK, to be accessible, BAM's Spark instance needs be started in
>>>>>>>> the clustering mode. Then any external client also can use the BAM's 
>>>>>>>> spark,
>>>>>>>> through the "spark.master" URL, which is "spark://localhost:7077" by
>>>>>>>> default. (Refer attached image "
>>>>>>>> external-spark-cluster.png").
>>>>>>>>
>>>>>>>> I'm not sure whether this feature (clustering Spark) is currently
>>>>>>>> supported by BAM (in M1 release)?
>>>>>>>>
>>>>>>>> If so, whats the way forward?
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Supun
>>>>>>>>
>>>>>>>> --
>>>>>>>> *Supun Sethunga*
>>>>>>>> Software Engineer
>>>>>>>> WSO2, Inc.
>>>>>>>> lean | enterprise | middleware
>>>>>>>> Mobile : +94 716546324
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> *Niranda Perera*
>>>>>>> Software Engineer, WSO2 Inc.
>>>>>>> Mobile: +94-71-554-8430
>>>>>>> Twitter: @n1r44 <https://twitter.com/N1R44>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>> *Supun Sethunga*
>>>>>> Software Engineer
>>>>>> WSO2, Inc.
>>>>>> lean | enterprise | middleware
>>>>>> Mobile : +94 716546324
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> *Supun Sethunga*
>>>>> Software Engineer
>>>>> WSO2, Inc.
>>>>> lean | enterprise | middleware
>>>>> Mobile : +94 716546324
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> *Niranda Perera*
>>>> Software Engineer, WSO2 Inc.
>>>> Mobile: +94-71-554-8430
>>>> Twitter: @n1r44 <https://twitter.com/N1R44>
>>>>
>>>
>>>
>>>
>>> --
>>> *Anjana Fernando*
>>> Senior Technical Lead
>>> WSO2 Inc. | http://wso2.com
>>> lean . enterprise . middleware
>>>
>>
>>
>>
>> --
>> *Niranda Perera*
>> Software Engineer, WSO2 Inc.
>> Mobile: +94-71-554-8430
>> Twitter: @n1r44 <https://twitter.com/N1R44>
>>
>
>
>
> --
> *Supun Sethunga*
> Software Engineer
> WSO2, Inc.
> http://wso2.com/
> lean | enterprise | middleware
> Mobile : +94 716546324
>
_______________________________________________
Dev mailing list
[email protected]
http://wso2.org/cgi-bin/mailman/listinfo/dev

Reply via email to