hadoop-yarn-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Charles Wright <charliewri...@live.ca>
Subject Negative Values for Pending Cores in QueueMetrics
Date Thu, 04 Feb 2016 02:39:04 GMT
I am trying to monitor cluster metrics using QueueMetrics and I am 
getting negative values for pending cores and pending memory. Is this a 
known issue?

What I am trying to do is send a "cluster pressure" notification to 
applications running the cluster so they can release any "extra" 
resources that they may have.

I have defined "cluster pressure" as:
(allocated containers + pending containers) / total available containers > 1

However, I cannot get it to work with negative pending cores/pending 
memory values.


View raw message