Wait until current task is "complete" and then accept no work

["DiGiorgio, Mr. Rinaldo S." <rd...@pace.edu>]

HI,

	One of the challenges I had with AWS was the inability to request that the automated scale up and scale down services wait for a system to reach a state where it was ok to perform that action. The AWS seemed to assume stateless, which is the simple use case. Can I request that a current working agent not accept new work when using a Mesos cluster?

Rinald

Re: Wait until current task is "complete" and then accept no work

[David Greenberg <ds...@gmail.com>]

There are a few ways to do this. One way is to use the maintenance API. Its
page does a good job of explaining how to use it:
http://mesos.apache.org/documentation/latest/maintenance/

Another approach is to use Satellite, Two Sigma's Mesos monitoring system.
Satellite would do the same thing as the Mesos API, but also monitor the
agent and generate an event when the agent is drained, so that you can run
JVM code within Satellite, or trigger an event in some other service or
system, like telling AWS to shutdown the VM.
https://github.com/twosigma/satellite

On Thu, May 5, 2016 at 9:23 AM DiGiorgio, Mr. Rinaldo S. <
rdigiorgio@pace.edu> wrote:

> HI,
>
>         One of the challenges I had with AWS was the inability to request
> that the automated scale up and scale down services wait for a system to
> reach a state where it was ok to perform that action. The AWS seemed to
> assume stateless, which is the simple use case. Can I request that a
> current working agent not accept new work when using a Mesos cluster?
>
> Rinald