> On Feb. 17, 2015, 6:49 p.m., Maxim Khutornenko wrote:
> > include/mesos/mesos.proto, lines 427-429
> > <https://reviews.apache.org/r/30873/diff/2/?file=864004#file864004line427>
> >
> > Just curious, any particular reason for the spread in tag IDs here?
>
> Michael Park wrote:
> Hi Maxim, the reason for this is **Remark 1** in the **Description**: "In `ReserverType`,
0 and 2 are left out for `NONE=0` and `OPERATOR=2` which we plan to introduce later on."
>
> Maxim Khutornenko wrote:
> Ah, thanks, makes sense now! I should've paid more attention to description :) BTW,
I really like the ReserverType, much more intuitive.
I really liked it initially as well, but it doesn't quite work well if we want to allow an
authorized framework with permissions to make operator level reservations. I think the name
ultimately needs to be tied to the level of reservation rather than the reserver. Similar
to a Linux environment where there are `SYSTEM` and `USER` level configs and an authorized
user with the right permissions is allowed to modify `SYSTEM` level configs but by default
a regular user can only change `USER` level configs. Anyway, the naming for these are still
in flight and hence the reason for this patch being marked `[WIP]`.
I'll create a JIRA ticket for this to capture essentially what I said here and also to get
some suggestions from the community :)
- Michael
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/30873/#review72756
-----------------------------------------------------------
On Feb. 14, 2015, 12:28 a.m., Michael Park wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/30873/
> -----------------------------------------------------------
>
> (Updated Feb. 14, 2015, 12:28 a.m.)
>
>
> Review request for mesos.
>
>
> Repository: mesos
>
>
> Description
> -------
>
> ## Overview
>
> We'll shortly be allowing operators to dynamically change the reservation configurations
of slaves as well. At which point the words `STATIC` and `DYNAMIC` are no longer accurate
since the `STATIC` reservations would also be dynamic. We change the terminology here to indicate
the source of the reservation, the reserver.
>
> ## Changes
>
> * `ReservationType` => `ReserverType`
> * `STATIC` => `SLAVE`
> * `DYNAMIC` => `FRAMEWORK`
>
> ## Remark
>
> 1. In `ReserverType`, 0 and 2 are left out for `NONE=0` and `OPERATOR=2` which we plan
to introduce later on.
> 2. This is marked WIP (Work in Progress) since the design of this hasn't been fully approved
yet.
> This chain of review requests are essentially a prototype.
>
>
> Diffs
> -----
>
> include/mesos/mesos.proto 6ece3ea30f9689db435b330422f8dfc08816316a
> src/master/allocator/allocator.hpp c2461a32e18dbafe637f37180f86bfa2e1d78735
> src/master/master.hpp 6a39df04514c756415354fae66c5835ada191c52
> src/messages/messages.proto 58484ae45071a80afd2b11803dd66a88f88ad9ed
> src/slave/slave.cpp ec7ec1356e745bb07484ae1755c9183b038043b3
>
> Diff: https://reviews.apache.org/r/30873/diff/
>
>
> Testing
> -------
>
> make check
>
>
> Thanks,
>
> Michael Park
>
>
|