You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@chukwa.apache.org by Bill Graham <bi...@gmail.com> on 2011/02/02 06:24:27 UTC
Re: Experiment with trunk code
We're not ready to migrate to HBase for our Chukwa processing at this
point. This is mainly because the current Demux approach is working
well for us and we aren't running HBase in production yet. We'd want
to run 0.5.0 using the current Demuxer and data processors.
On Sun, Jan 30, 2011 at 10:47 AM, Eric Yang <er...@gmail.com> wrote:
> Hi all,
>
> How are we doing in adopting HBase as part of required Chukwa
> component? I wonder if anyone has comment or feedback?
>
> For releasing Chukwa 0.5, we need 2 more features implemented.
> CHUKWA-567, and CHUKWA-576. One is dealing with down sampling, and
> another one is creating UDF to improve counter and gauge metrics
> calculation. Is anyone interested in taking on those tasks?
>
> I am working on creating rpm target for hadoop and pig for ASF. This
> should improve deployment experience for the end user.
>
> regards,
> Eric
>
Re: Experiment with trunk code
Posted by Eric Yang <ey...@yahoo-inc.com>.
Those features should continue to work as they are. I think you are fine.
Regards,
Eric
On 2/1/11 10:55 PM, "Bill Graham" <bi...@gmail.com> wrote:
No, we're not using MetricsDataLoader or any of the HICC/MySQL
functionality so we should be all good. We use Demux for 5 minute
aggregation and the hourly/daily roll-up processes with our own
summary jobs.
On Tue, Feb 1, 2011 at 10:29 PM, Eric Yang <er...@gmail.com> wrote:
> Hi Bill,
>
> Chukwa 0.5 still supports demux by mapreduce. However, the database
> code has been removed. Does your deployment depends on
> MetricsDataLoader?
>
> regards,
> Eric
>
> On Tue, Feb 1, 2011 at 9:24 PM, Bill Graham <bi...@gmail.com> wrote:
>> We're not ready to migrate to HBase for our Chukwa processing at this
>> point. This is mainly because the current Demux approach is working
>> well for us and we aren't running HBase in production yet. We'd want
>> to run 0.5.0 using the current Demuxer and data processors.
>>
>>
>> On Sun, Jan 30, 2011 at 10:47 AM, Eric Yang <er...@gmail.com> wrote:
>>> Hi all,
>>>
>>> How are we doing in adopting HBase as part of required Chukwa
>>> component? I wonder if anyone has comment or feedback?
>>>
>>> For releasing Chukwa 0.5, we need 2 more features implemented.
>>> CHUKWA-567, and CHUKWA-576. One is dealing with down sampling, and
>>> another one is creating UDF to improve counter and gauge metrics
>>> calculation. Is anyone interested in taking on those tasks?
>>>
>>> I am working on creating rpm target for hadoop and pig for ASF. This
>>> should improve deployment experience for the end user.
>>>
>>> regards,
>>> Eric
>>>
>>
>
Re: Experiment with trunk code
Posted by Bill Graham <bi...@gmail.com>.
No, we're not using MetricsDataLoader or any of the HICC/MySQL
functionality so we should be all good. We use Demux for 5 minute
aggregation and the hourly/daily roll-up processes with our own
summary jobs.
On Tue, Feb 1, 2011 at 10:29 PM, Eric Yang <er...@gmail.com> wrote:
> Hi Bill,
>
> Chukwa 0.5 still supports demux by mapreduce. However, the database
> code has been removed. Does your deployment depends on
> MetricsDataLoader?
>
> regards,
> Eric
>
> On Tue, Feb 1, 2011 at 9:24 PM, Bill Graham <bi...@gmail.com> wrote:
>> We're not ready to migrate to HBase for our Chukwa processing at this
>> point. This is mainly because the current Demux approach is working
>> well for us and we aren't running HBase in production yet. We'd want
>> to run 0.5.0 using the current Demuxer and data processors.
>>
>>
>> On Sun, Jan 30, 2011 at 10:47 AM, Eric Yang <er...@gmail.com> wrote:
>>> Hi all,
>>>
>>> How are we doing in adopting HBase as part of required Chukwa
>>> component? I wonder if anyone has comment or feedback?
>>>
>>> For releasing Chukwa 0.5, we need 2 more features implemented.
>>> CHUKWA-567, and CHUKWA-576. One is dealing with down sampling, and
>>> another one is creating UDF to improve counter and gauge metrics
>>> calculation. Is anyone interested in taking on those tasks?
>>>
>>> I am working on creating rpm target for hadoop and pig for ASF. This
>>> should improve deployment experience for the end user.
>>>
>>> regards,
>>> Eric
>>>
>>
>
Re: Experiment with trunk code
Posted by Eric Yang <er...@gmail.com>.
Hi Bill,
Chukwa 0.5 still supports demux by mapreduce. However, the database
code has been removed. Does your deployment depends on
MetricsDataLoader?
regards,
Eric
On Tue, Feb 1, 2011 at 9:24 PM, Bill Graham <bi...@gmail.com> wrote:
> We're not ready to migrate to HBase for our Chukwa processing at this
> point. This is mainly because the current Demux approach is working
> well for us and we aren't running HBase in production yet. We'd want
> to run 0.5.0 using the current Demuxer and data processors.
>
>
> On Sun, Jan 30, 2011 at 10:47 AM, Eric Yang <er...@gmail.com> wrote:
>> Hi all,
>>
>> How are we doing in adopting HBase as part of required Chukwa
>> component? I wonder if anyone has comment or feedback?
>>
>> For releasing Chukwa 0.5, we need 2 more features implemented.
>> CHUKWA-567, and CHUKWA-576. One is dealing with down sampling, and
>> another one is creating UDF to improve counter and gauge metrics
>> calculation. Is anyone interested in taking on those tasks?
>>
>> I am working on creating rpm target for hadoop and pig for ASF. This
>> should improve deployment experience for the end user.
>>
>> regards,
>> Eric
>>
>