----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/60198/ -----------------------------------------------------------
Review request for Ambari, Aravindan Vijayan and Sid Wagle. Bugs: AMBARI-21279 https://issues.apache.org/jira/browse/AMBARI-21279 Repository: ambari Description ------- If monitor aggregation is enabled we do not do collector side 5 minutes aggregation. If a monitor is down we do not get metrics from that host. Metrics collector should check that metrics from a host are available per minute but aggregates are not. This could be based on liveliness check of a monitor. More like a heartbeat ping. When we loose ping for > 1 min < 5 min we can aggregate metrics for that time period and hostId. Sinks and monitor should switch back to collector reporting state if local aggregation app is not available. Diffs ----- ambari-metrics/ambari-metrics-timelineservice/src/main/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/HBaseTimelineMetricStore.java f984253 ambari-metrics/ambari-metrics-timelineservice/src/main/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricAggregatorFactory.java 2eb3553 ambari-metrics/ambari-metrics-timelineservice/src/main/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/TimelineMetricHostConditionalAggregator.java PRE-CREATION ambari-metrics/ambari-metrics-timelineservice/src/main/java/org/apache/hadoop/yarn/server/applicationhistoryservice/metrics/timeline/aggregators/v2/TimelineMetricHostConditionalAggregator.java PRE-CREATION Diff: https://reviews.apache.org/r/60198/diff/1/ Testing ------- In progress Thanks, Dmytro Sen