mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael Park" <mcyp...@gmail.com>
Subject Re: Review Request 32982: Added reservation user guide.
Date Tue, 30 Jun 2015 04:57:11 GMT


> On June 29, 2015, 11:43 p.m., Adam B wrote:
> > docs/reservation.md, line 108
> > <https://reviews.apache.org/r/32982/diff/12/?file=994059#file994059line108>
> >
> >     I know that we'll want to explicitly specify the role to support future multi-role
frameworks, but why does this need to be included now?

There were 2 motivations for this. First is what you said, the second was that we can enforce
the field to be `required` this way.


> On June 29, 2015, 11:43 p.m., Adam B wrote:
> > docs/reservation.md, lines 109-110
> > <https://reviews.apache.org/r/32982/diff/12/?file=994059#file994059line109>
> >
> >     Shouldn't the master be able to determine the principal that this frameworkId
is currently authenticated as? Why even allow the framework to specify a different principal
if you're just going to Error?

Hm, that's true. This was another situation where I wanted to keep  `ReservationInfo.principal`
as `required`.


> On June 29, 2015, 11:43 p.m., Adam B wrote:
> > docs/reservation.md, line 196
> > <https://reviews.apache.org/r/32982/diff/12/?file=994059#file994059line196>
> >
> >     Could also only unreserve a subset, right?

Yeah, that's true. I didn't really mention that part.


> On June 29, 2015, 11:43 p.m., Adam B wrote:
> > docs/reservation.md, line 309
> > <https://reviews.apache.org/r/32982/diff/12/?file=994059#file994059line309>
> >
> >     How can there be insufficient resources to unreserve?

The simplest case would be: if we have 4 dynamically reserved cpus and the request is to unreserve
6 dynamically reserved cpus. We don't simply unreserve the portion we can. A slightly more
complicated case would be: if we have 4 dynamically reserved cpus and the request is to unreserve
4 dynamically reserved cpus, but 2 of them are currently being used. As we don't currently
pre-empt tasks, it results in a `Conflict` for us currently.


- Michael


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/32982/#review89824
-----------------------------------------------------------


On June 28, 2015, 3:33 a.m., Michael Park wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/32982/
> -----------------------------------------------------------
> 
> (Updated June 28, 2015, 3:33 a.m.)
> 
> 
> Review request for mesos, Alexander Rukletsov, Jie Yu, and Timothy Chen.
> 
> 
> Bugs: MESOS-2205
>     https://issues.apache.org/jira/browse/MESOS-2205
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> The Github rendered version is available [here]( https://github.com/mesosphere/mesos/blob/user-docs/docs/reservation.md)
> 
> 
> Diffs
> -----
> 
>   docs/reservation.md PRE-CREATION 
> 
> Diff: https://reviews.apache.org/r/32982/diff/
> 
> 
> Testing
> -------
> 
> Documentation.
> 
> 
> Thanks,
> 
> Michael Park
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message