> On Jan. 15, 2014, 2:04 a.m., Benjamin Hindman wrote:
> > Does it still make sense to continue to collect statistics at an interval 
> > or only drive it completely by the REST endpoint? Perhaps you want other 
> > endpoints, like archive or time series?
> 
> Ben Mahler wrote:
>     The only reason we periodically collected was to eventually show 
> monitoring history in the webui. If we wanted to show history and the 
> collection was driven by the REST endpoint, we would not have any historical 
> data to show, unless something is hitting the endpoint periodically.
>     
>     Since we're not close to showing monitoring history in the webui, I would 
> be ok with removing the periodic collection and the historical monitoring 
> information until we know how we're going to use this data in the webui. What 
> do you think?
> 
> Ben Mahler wrote:
>     These are the two choices I see for this:
>     
>     1. Removing historical data, and driving collection only by the endpoint. 
> Or,
>     2. Keeping historical data, and doing our own periodic collection.

I prefer 2 because 1 is not reliable if there is a network problem.


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/13605/#review31842
-----------------------------------------------------------


On Jan. 15, 2014, 2:06 a.m., Ben Mahler wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/13605/
> -----------------------------------------------------------
> 
> (Updated Jan. 15, 2014, 2:06 a.m.)
> 
> 
> Review request for mesos, Benjamin Hindman and Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> This alters statistics.json to return instantaneous resource consumption 
> information.
> 
> Previously, one could receive information that is stale by up to 5 seconds 
> (or the RESOURCE_MONITORING_INTERVAL). This allows one to build Top like 
> utilities using the monitoring endpoint, this is done by the subsequent 
> change in this chain of reviews.
> 
> 
> Diffs
> -----
> 
>   src/slave/monitor.hpp 52568ad8ec566f7cf36c249c76d798d44eacb578 
>   src/slave/monitor.cpp a931c4f35a8793c66ee03de82f0e0a21b92f8ffa 
>   src/tests/monitor_tests.cpp a341893b16fbe502fa32704fcd1f3f85519ad253 
> 
> Diff: https://reviews.apache.org/r/13605/diff/
> 
> 
> Testing
> -------
> 
> Split the monitoring tests into two tests:
> 
> 1. Tests that verify the periodic collection is working correctly.
> 2. Tests that use the instantaneous statistics.json endpoint.
> 
> 
> Thanks,
> 
> Ben Mahler
> 
>

Reply via email to