Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
MapReduce >> mail # user >> Best way to collect Hadoop logs across cluster


Copy link to this message
-
Re: Best way to collect Hadoop logs across cluster
Actually the problem is not simple. Based on these problems, there are
three companies working on them:
- Loggly:
  http://loggly.com/

  Loggly is a part of the Amazon Marketplace:
  https://aws.amazon.com/solution-providers/isv/loggly

- Pageduty:
  https://papertrailapp.com/

  How to do it:

http://help.papertrailapp.com/discussions/questions/1767-logging-to-heroku-papertrail-addon-from-an-app-on-external-server

 http://help.papertrailapp.com/kb/how-it-works/permanent-log-archives

http://help.papertrailapp.com/kb/analytics/log-analytics-with-hadoop-and-hive
- Splunk, with its Cloud platform Splunk Storm:
  https://www.splunkstorm.com/
With any of these applications, you can collect permanently or based on
schedule all generated logs from Amazon EC2, which is the base service for
EMR and the others.

Best wishes.

2013/4/19 Mark Kerzner <[EMAIL PROTECTED]>

> So you are saying, the problem is very simple. Just before you destroy the
> cluster, simply collect the logs to S3. Anyway, I only need them after I
> have completed with a specific computation. So I don't need any special
> requirements.
>
> In regular permanent clusters, is there something that allows you to see
> all logs in one place?
>
> Thank you,
> Mark
>
>
> On Thu, Apr 18, 2013 at 11:51 PM, Marcos Luis Ortiz Valmaseda <
> [EMAIL PROTECTED]> wrote:
>
>> When you destroy an EC2 instance, the correct behavior is to erase all
>> data.
>> Why don't you create a service to collect the logs directly to a S3
>> bucket in real-time or in a batch of 5 mins?
>>
>>
>> 2013/4/18 Mark Kerzner <[EMAIL PROTECTED]>
>>
>>> Hi,
>>>
>>> my clusters are on EC2, and they disappear after the cluster's instances
>>> are destroyed. What is the best practice to collect the logs for later
>>> storage?
>>>
>>> EC2 does exactly that with their EMR, how do they do it?
>>>
>>> Thank you,
>>> Mark
>>>
>>
>>
>>
>> --
>> Marcos Ortiz Valmaseda,
>> *Data-Driven Product Manager* at PDVSA
>> *Blog*: http://dataddict.wordpress.com/
>> *LinkedIn: *http://www.linkedin.com/in/marcosluis2186
>> *Twitter*: @marcosluis2186 <http://twitter.com/marcosluis2186>
>>
>
>
--
Marcos Ortiz Valmaseda,
*Data-Driven Product Manager* at PDVSA
*Blog*: http://dataddict.wordpress.com/
*LinkedIn: *http://www.linkedin.com/in/marcosluis2186
*Twitter*: @marcosluis2186 <http://twitter.com/marcosluis2186>