Thanks for those links. I'm using the DEA for now and it looks like
nothing has changed there for quite some time (Apr 2014). I'll see if I
can debug a bit more to see what's happening.

Thanks,

Dan

On Mon, Oct 5, 2015 at 4:36 PM, Matthew Sykes <matthew.sykes(a)gmail.com>
wrote:

In the DEA:
https://github.com/cloudfoundry/dea_ng/blob/4b1a50ae5598b0c70cb3e5895ed800e0cff37722/lib/dea/stat_collector.rb#L93

In Diego/Garden:

https://github.com/cloudfoundry-incubator/garden-linux/blob/master/linux_container/metrics.go#L129

If you're using the DEA, stats come back in response to the `find.droplet`
message while in Diego they get updated on an interval.

On Mon, Oct 5, 2015 at 4:20 PM, Daniel Mikusa <dmikusa(a)pivotal.io> wrote:

Has something changed recently with the output of `cf app` and how it
reports memory usage?

I was deploying apps to CF v219 and noticed that the output of `cf app`
is not what I'm used to seeing.

I'm deploying a Java app with a 512M memory limit. The Java build pack
is setting the initial heap and metaspace sizes to be 373M and 64M
(-XX:MaxMetaspaceSize=64M -Xss995K -Xmx382293K -Xms382293K
-XX:MetaspaceSize=64M), which is a total of 437M. Since that's just the
heap and metaspace, I'd expect the app to start out using at least that
much memory.

The report from `cf app` is showing 366.3M of 512M.

Does anyone know how this is presently being calculated / have a link to
the source code where this is being calculated?

Thanks,

Dan

--
Matthew Sykes
matthew.sykes(a)gmail.com