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

Switch to Threaded View
Flume >> mail # user >> Configuring failover sync in Flume-ng


Copy link to this message
-
Re: Configuring failover sync in Flume-ng
Yes, why not luring the wiki :)
https://cwiki.apache.org/confluence/display/FLUME/Home

Stable: http://www.apache.org/dyn/closer.cgi/flume/1.3.0
Just grab the tar.gz - or use 1.2.0 as deb-package, I suspect the balancing stuff is inside too.

Best,
 Alex

On Dec 19, 2012, at 2:55 PM, Abhijeet Pathak <[EMAIL PROTECTED]> wrote:

> I am running Ubuntu 12.04 server edition.
> The repo has flume 1.2 release available.
> How should I install 1.3? Using source?
> Is there a stable release available?
>
> Thanks,
> Abhijeet
>
> -----Original Message-----
> From: Alexander Alten-Lorenz [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, December 19, 2012 4:33 PM
> To: [EMAIL PROTECTED]
> Subject: Re: Configuring failover sync in Flume-ng
>
> Oh, yes. You should upgrade into 1.3.0, which was released last week. Wasn't clear for me that you're running on 1.1.0
>
> Do you have a change to upgrade into the newest release?
>
> Best,
> Alex
>
> On Dec 19, 2012, at 11:51 AM, Abhijeet Pathak <[EMAIL PROTECTED]> wrote:
>
>> It's still not working.
>>
>> I read somewhere that failover support was implemented after Jan 2012.
>> I've flume 1.1.0+121-1.cdh4.0.1.p0.1~precise-cdh4.0.1 installed.
>>
>> Can that be a reason for it not working?
>>
>> Thanks,
>> Abhijeet
>>
>> -----Original Message-----
>> From: Alexander Alten-Lorenz [mailto:[EMAIL PROTECTED]]
>> Sent: Wednesday, December 19, 2012 2:52 PM
>> To: [EMAIL PROTECTED]
>> Subject: Re: Configuring failover sync in Flume-ng
>>
>> YOUR_NAME.sinks = HOSTSINK-1 HOSTSINK-2
>>
>> HOSTSINK-1.type>> HOSTSINK-1.bind=HOSTNAME-1
>> HOSTSINK-1.port=needed by JDBC?
>> HOSTSINK-1.channels=YOUR_CHANNELS
>>
>> the same for the other HOSTSINK and deploy the config to the hosts and startup. Should work.
>>
>> Best,
>> Alex
>>
>> On Dec 19, 2012, at 10:01 AM, Abhijeet Pathak <[EMAIL PROTECTED]> wrote:
>>
>>> Yes, I saw that.
>>>
>>> But I'm not sure how to tell flume that the failover sink is running on "another" machine.
>>>
>>> -----Original Message-----
>>> From: Alexander Lorenz [mailto:[EMAIL PROTECTED]]
>>> Sent: Wednesday, December 19, 2012 1:59 PM
>>> To: [EMAIL PROTECTED]
>>> Subject: Re: Configuring failover sync in Flume-ng
>>>
>>> Hi,
>>>
>>> Failover:
>>> http://flume.apache.org/FlumeUserGuide.html#failover-sink-processor
>>>
>>> Just define more as one JDBC channel and add them to the failover config (see the examples).
>>>
>>> Best
>>> - Alex
>>>
>>> --
>>> Alexander Alten-Lorenz
>>> http://mapredit.blogspot.com
>>> German Hadoop LinkedIn Group: http://goo.gl/N8pCF
>>>
>>> On Dec 19, 2012, at 9:23 AM, Abhijeet Pathak <[EMAIL PROTECTED]> wrote:
>>>
>>>> Hi,
>>>>
>>>> I've working setup of flume-ng agent (1 source [tail], 1 channel[jdbc], 1 sink [AsyncHbase]) on one server.
>>>>
>>>> However, I am not able to find documentation on how to configure failover sinks for such scenario.
>>>>
>>>> I want to run same sink as backup/failover sink on different server.
>>>> How can that be configured?
>>>>
>>>> Thanks,
>>>> Abhijeet
>>>>
>>>
>>
>