[
https://issues.apache.org/jira/browse/AMBARI-16695?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15285739#comment-15285739
]
Siddharth Wagle commented on AMBARI-16695:
------------------------------------------
|| Alert || Existing Default thresholds (warn/crit) || Minimum value ||
| increase_nn_heap_usage_daily | 20 / 50 | 100 MB |
| increase_nn_heap_usage_weekly | 20 / 50 | 1 GB |
| namenode_increase_in_storage_capacity_usage_daily | 30 / 50 | 100 MB |
| namenode_increase_in_storage_capacity_usage_weekly | 10 / 20 | 1 GB |
Effectively we will not alert if user adds < 100 MB every day and fills up the
cluster. I still think it make common sense to have these.
> HDFS Alerts: add minimum values to AMS alerts
> ---------------------------------------------
>
> Key: AMBARI-16695
> URL: https://issues.apache.org/jira/browse/AMBARI-16695
> Project: Ambari
> Issue Type: Improvement
> Reporter: Siddharth Wagle
> Assignee: Siddharth Wagle
>
> There are new HDFS alerts that watch growth rates. Some (like RPC) have
> "minimum" values, meaning we ignore growth until we are past a certain value
> (like latency in seconds).
> There are a few other alerts that I think also need minimums. Or maybe need
> higher warning thresholds. Or both...
> ======
> NameNode Heap Usage (Daily) increase_nn_heap_usage_daily
> I'm getting these WARNINGS with a Heap of < 100MB.
> {code}
> The variance for this alert is 25MB which is 30% of the 82MB average (16MB is
> the limit).
> {code}
> ====
> HDFS Storage Capacity Usage (Weekly)
> namenode_increase_in_storage_capacity_usage_weekly
> I'm getting these WARNINGS when only using such a small amount of storage...
> {code}
> The variance for this alert is 21,328B which is 20% of the 107,758B average
> (10,776B is the limit)
> {code}
> Here is what is used...
> {code}
> Disk Usage (DFS Used) 112.0 KB / 428.1 GB (0.00%)
> Disk Usage (Non DFS Used) 32.0 GB / 428.1 GB (7.47%)
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
