You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by "Tom Beerbower (JIRA)" <ji...@apache.org> on 2012/10/25 16:35:11 UTC
[jira] [Updated] (AMBARI-911) Implement an efficient way to provide
Ganglia data for heatmap and other cluster visualization tools
[ https://issues.apache.org/jira/browse/AMBARI-911?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Tom Beerbower updated AMBARI-911:
---------------------------------
Description:
We need to be able to render cluster heatmaps that represents the value of a specified metric (or a combination of it) for all hosts in the cluster.
See the attached screenshot.
Rendering a cluster heatmap requires retrieving a metric (or a set of metrics) across all hosts in the cluster (current or any point-in-time in the past for which the data points are available).
We are using Ganglia Web's graph.php to generate JSON for data points of interest. However, this mechanism is not suited for cluster heatmap rendering purposes as it would require N calls to Ganglia Web where N is the number of hosts in the cluster.
We may need to implement a custom .php that we can deploy on Ganglia Web to work directly with RRD in order for us to provide this data.
Also, we need to create appropriate API endpoints to expose such data.
The API should accept a list of metrics, a list of hostnames, a timestamp (probably in seconds since UNIX epoch) or a pair of timestamps for specifying a range, and "steps" to define granularity of the output data points.
Also, the custom .php would ideally accept a number of data points to aggregate the time range over; this way, we can not only show point-in-time metrics, but also aggregated metrics over the X hours prior to the specified timestamp. This would be useful for showing something like: "what was the average CPU usage for all hosts in the cluster for the past 2 weeks?"
was:
In Bimota, we need to be able to render cluster heatmaps that represents the value of a specified metric (or a combination of it) for all hosts in the cluster.
See the attached screenshot.
Rendering a cluster heatmap requires retrieving a metric (or a set of metrics) across all hosts in the cluster (current or any point-in-time in the past for which the data points are available).
We are using Ganglia Web's graph.php to generate JSON for data points of interest. However, this mechanism is not suited for cluster heatmap rendering purposes as it would require N calls to Ganglia Web where N is the number of hosts in the cluster.
We may need to implement a custom .php that we can deploy on Ganglia Web to work directly with RRD in order for us to provide this data.
Also, we need to create appropriate API endpoints to expose such data.
The API should accept a list of metrics, a list of hostnames, a timestamp (probably in seconds since UNIX epoch) or a pair of timestamps for specifying a range, and "steps" to define granularity of the output data points.
Also, the custom .php would ideally accept a number of data points to aggregate the time range over; this way, we can not only show point-in-time metrics, but also aggregated metrics over the X hours prior to the specified timestamp. This would be useful for showing something like: "what was the average CPU usage for all hosts in the cluster for the past 2 weeks?"
> Implement an efficient way to provide Ganglia data for heatmap and other cluster visualization tools
> ----------------------------------------------------------------------------------------------------
>
> Key: AMBARI-911
> URL: https://issues.apache.org/jira/browse/AMBARI-911
> Project: Ambari
> Issue Type: Sub-task
> Reporter: Tom Beerbower
> Assignee: Tom Beerbower
>
> We need to be able to render cluster heatmaps that represents the value of a specified metric (or a combination of it) for all hosts in the cluster.
> See the attached screenshot.
> Rendering a cluster heatmap requires retrieving a metric (or a set of metrics) across all hosts in the cluster (current or any point-in-time in the past for which the data points are available).
> We are using Ganglia Web's graph.php to generate JSON for data points of interest. However, this mechanism is not suited for cluster heatmap rendering purposes as it would require N calls to Ganglia Web where N is the number of hosts in the cluster.
> We may need to implement a custom .php that we can deploy on Ganglia Web to work directly with RRD in order for us to provide this data.
> Also, we need to create appropriate API endpoints to expose such data.
> The API should accept a list of metrics, a list of hostnames, a timestamp (probably in seconds since UNIX epoch) or a pair of timestamps for specifying a range, and "steps" to define granularity of the output data points.
> Also, the custom .php would ideally accept a number of data points to aggregate the time range over; this way, we can not only show point-in-time metrics, but also aggregated metrics over the X hours prior to the specified timestamp. This would be useful for showing something like: "what was the average CPU usage for all hosts in the cluster for the past 2 weeks?"
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira