[
https://issues.apache.org/jira/browse/MESOS-4666?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15146406#comment-15146406
]
Anindya Sinha edited comment on MESOS-4666 at 2/15/16 7:53 PM:
---------------------------------------------------------------
> For 2) Why do you want exclusive access? Is it because the current isolation
> (resource and security) in Mesos is not
> good/compliant enough or something else?
No. The reason is due to the existing architecture of the service that we wish
to deploy. The constraints for deployment of this service requires n instances
of app1 to be colocated with 1 instance of app2.
> Anyway, the ability to somehow dynamically reserve a whole slave is an
> interesting use case. For that we might have
> to expose the total amount of resources, without exposing the reservation
> information.
But exposing total resource count which does not include the resource count of
any reserved resources of other roles would mean that the info received by the
framework is not a guarantee of the actual total resources which means
exclusivity may or may not not be possible (esp in a cluster with varying sizes
of agent nodes in terms of resources).
was (Author: anindya.sinha):
> For 2) Why do you want exclusive access? Is it because the current isolation
> (resource and security) in Mesos is not
> good/compliant enough or something else?
No. The reason is due to the existing architecture of the service that we wish
to deploy. The constraints for deployment of this service requires n instances
of app1 to be colocated with 1 instance of app2.
> Anyway, the ability to somehow dynamically reserve a whole slave is an
> interesting use case. For that we might have
> to expose the total amount of resources, without exposing the reservation
> information.
But exposing total resources which does not include reservation info of other
roles would mean that the info received by the framework is not a guarantee of
the actual total resources which means exclusivity may not be possible (esp in
a cluster with varying sizes of agent nodes in terms of resources).
Although indirectly it is the same, would instead of total_resources as an
actual value, exposing a %availability a better option? ie. if a slave is
registered with {noformat} cpus(*): 6;cpus(role1): 2;mem(*):8192;mem(role1):
8192;disk(*): 102400;disk(role1): 61440 {noformat}, we send % of each resource
as follows:
- If no resource has been allocated yet, we return Offer::total_resources =
{noformat} cpus(*): 1.0;mem(*): 1.0;disk(*): 1.0 {noformat}
- If say 2 cpus, 6144MB mem and 30720MB disk have already been allocated, then
we return Offer::total_resources = {noformat} cpus(*): 0.75;mem(*):
0.625;disk(*): 0.8125 {noformat} and so on.
> Expose total resources of a slave in offer for scheduling decisions
> -------------------------------------------------------------------
>
> Key: MESOS-4666
> URL: https://issues.apache.org/jira/browse/MESOS-4666
> Project: Mesos
> Issue Type: Improvement
> Components: general
> Affects Versions: 0.25.0
> Reporter: Anindya Sinha
> Assignee: Anindya Sinha
> Priority: Minor
>
> To effectively schedule certain class of tasks, the scheduler might need to
> know not only the available resources (as exposed currently) but also the
> maximum resources available on that slave. This is specifically true for
> clusters having different configurations of the slave nodes in terms of
> resources such as cpu, memory, disk, etc.
> Certain class of tasks might have a need to be scheduled on the same slave
> (esp needing shared persistent volumes, MESOS-3421). Instead of dedicating a
> slave to a framework, the framework can make a very good determination if it
> had exposure to both available as well as total resources.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
