ambari-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sid Wagle" <swa...@hortonworks.com>
Subject Re: Review Request 39758: Add normalizer properties for AMS Regions
Date Thu, 29 Oct 2015 17:32:47 GMT


> On Oct. 29, 2015, 4:01 p.m., Sid Wagle wrote:
> > ambari-metrics/ambari-metrics-timelineservice/conf/unix/ambari-metrics-collector,
line 34
> > <https://reviews.apache.org/r/39758/diff/2/?file=1112158#file1112158line34>
> >
> >     We should do a little more analysis of how expensive this is, my initial though
was only choose 3 tables : METRIC_RECORD, METRIC_RECORD_MINUTE and METRIC_AGGREGATE.
> >     
> >     Better if we can make this configurable, we can add an export t0 ams-env.
> 
> Dmytro Sen wrote:
>     I think there is no need to make it configurable. Other tables are just skipped by
the normalizer, they should have at least 3 regions to be considered for normalization. 
>     They are updated very rarely, so we can expect split/merge their regions once per
month or even less often.

I see, ok lets make sure document that on the wiki. I will drop this issue.


> On Oct. 29, 2015, 4:01 p.m., Sid Wagle wrote:
> > ambari-metrics/ambari-metrics-timelineservice/conf/unix/ambari-metrics-collector,
line 92
> > <https://reviews.apache.org/r/39758/diff/2/?file=1112158#file1112158line92>
> >
> >     This should have some timeout value.
> 
> Dmytro Sen wrote:
>     alter_async is executed asyncronously, in backgound, so start command execution time
isn't increased.
>     We don't wait, that's a reason why we can't log if the altering table succeeded or
not. If use syncronous alter command, collector start time can take up to 2 minutes.
>     
>     That's a workaround until normalization enabling supported by the Phoenix DDL

Very good, let me open a Jira for that to happen.


- Sid


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/39758/#review104438
-----------------------------------------------------------


On Oct. 29, 2015, 2:59 p.m., Dmytro Sen wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/39758/
> -----------------------------------------------------------
> 
> (Updated Oct. 29, 2015, 2:59 p.m.)
> 
> 
> Review request for Ambari, Myroslav Papirkovskyy and Sid Wagle.
> 
> 
> Bugs: AMBARI-13632
>     https://issues.apache.org/jira/browse/AMBARI-13632
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> Test out the normalizer on AMS Regions for tables: METRIC_RECORD, METRIC_RECORD_MINUTE
and METRIC_AGGREGATE
> Use the load simulator for these tests
> Objective is to find out how often to run normalizer as a function of (sinks/hosts)
> Make stack advisor changes as needed
> 
> 
> Diffs
> -----
> 
>   ambari-metrics/ambari-metrics-timelineservice/conf/unix/ambari-metrics-collector dd7ee22

>   ambari-server/src/main/resources/common-services/AMBARI_METRICS/0.1.0/configuration/ams-hbase-site.xml
17f9b81 
> 
> Diff: https://reviews.apache.org/r/39758/diff/
> 
> 
> Testing
> -------
> 
> Unit tests passed
> 
> 
> Thanks,
> 
> Dmytro Sen
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message