[
https://issues.apache.org/jira/browse/AMBARI-13863?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aravindan Vijayan updated AMBARI-13863:
---------------------------------------
Description:
Problem
On the Ambari dashboard page, when different time ranges are randomly chosen,
the metrics charts stop responding and an OutOfMemory is seen in the Ambari
server logs.
Bug
In the AMS cache layer, the logic to pad the cache entry map by looking at the
delta between the values present in the map. When there is only one data point,
it assumes a default pad interval of 15 seconds even if the actual data
precision could be in minutes/hours/days.
Fix
Change the padding interval logic to use the precision calculated using the
requested start and end time. If the precision is in SECONDS, the delta between
the values in the map are used.
was:
Problem
On the Ambari dashboard page, when different time ranges are randomly chosen,
the metrics charts stop responding and an OutOfMemory is seen in the Ambari
server logs.
Bug
In the AMS cache layer, the logic to update the cache entry map by removing
values outside the given requested time range eventually leads to memory leak.
Fix
Change the logic to explicitly by using the iterator.remove() to remove the
entries instead of using the map.clear() method.
> UI lags and OutOfMemoryError seen in Ambari server after dashboard metrics'
> time periods are changed.
> -----------------------------------------------------------------------------------------------------
>
> Key: AMBARI-13863
> URL: https://issues.apache.org/jira/browse/AMBARI-13863
> Project: Ambari
> Issue Type: Bug
> Components: ambari-metrics
> Affects Versions: 2.1.2
> Reporter: Aravindan Vijayan
> Assignee: Aravindan Vijayan
> Priority: Critical
> Fix For: 2.1.3
>
> Attachments: AMBARI-13863.patch
>
>
> Problem
> On the Ambari dashboard page, when different time ranges are randomly chosen,
> the metrics charts stop responding and an OutOfMemory is seen in the Ambari
> server logs.
> Bug
> In the AMS cache layer, the logic to pad the cache entry map by looking at
> the delta between the values present in the map. When there is only one data
> point, it assumes a default pad interval of 15 seconds even if the actual
> data precision could be in minutes/hours/days.
> Fix
> Change the padding interval logic to use the precision calculated using the
> requested start and end time. If the precision is in SECONDS, the delta
> between the values in the map are used.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
