ambari-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sid Wagle" <swa...@hortonworks.com>
Subject Review Request 38415: Kafka metrics result in skewed split point distribiution
Date Wed, 16 Sep 2015 00:18:08 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/38415/
-----------------------------------------------------------

Review request for Ambari, Alejandro Fernandez, Dmytro Sen, Myroslav Papirkovskyy, and Sumit
Mohanty.


Bugs: AMBARI-13108
    https://issues.apache.org/jira/browse/AMBARI-13108


Repository: ambari


Description
-------

Out of 2400 data points Kafka metrics take up 1100. In a 400 node cluster this results in
an imbalance among the load distribution on the Region Server.


Diffs
-----

  ambari-server/src/main/resources/common-services/AMBARI_METRICS/0.1.0/package/scripts/split_points.py
b8b38f9 

Diff: https://reviews.apache.org/r/38415/diff/


Testing
-------

Unit test pass.

Manual test results:

python2.6 split_points.py '{"hbase.regionserver.global.memstore.upperLimit":"0.35", "hbase.hregion.memstore.flush.size":"268435456"}'
'{"hbase_master_heapsize":"24576m","hbase_regionserver_heapsize":"8192m"}' 'AMBARI_METRICS,HDFS,YARN,HBASE,KAFKA,FLUME'

Split points for precision table : 30
Split points for aggregate table : 7


Thanks,

Sid Wagle


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