mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jiang Yan Xu <...@jxu.me>
Subject Re: Review Request 45961: Support sharing of resources through reference counting of resources.
Date Wed, 07 Sep 2016 19:23:31 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/45961/#review148064
-----------------------------------------------------------



Given the way this review has evolved how about the following summary for the commit?

```
Support sharing of persistent volumes via shared resources.

- One copy of a shared resource is made allocatable in each allocation 
  cycle. 
- All frameworks that are allocated the same shared resource are
  charged the scalar quantity of the resource in wDRF.
- Resource accounting for *individual* copies of shared resources is
  handled mostly in the same way as non-shared resources in the master 
  although they are grouped within the `Resources` abstraction for
  performance.
- DESTROY for shared persistent volumes is allowed only if there are
  no running or pending tasks using the volume. However, if the volume
  is in a pending offer to one or more frameworks, the master rescinds
  that offer before processing the DESTROY.
- Multiple tasks are allowed to be launched in the same ACCEPT call
  using the same shared resource. For this we update the allocator and
  sorter to allocate additional copies of these shared resources.
```

- Jiang Yan Xu


On Sept. 7, 2016, 10:56 a.m., Anindya Sinha wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/45961/
> -----------------------------------------------------------
> 
> (Updated Sept. 7, 2016, 10:56 a.m.)
> 
> 
> Review request for mesos, Benjamin Mahler, Joris Van Remoortere, Michael Park, and Jiang
Yan Xu.
> 
> 
> Bugs: MESOS-4431
>     https://issues.apache.org/jira/browse/MESOS-4431
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> o Each shared resource is accouted via its share count. This count is
>   updated based on the resource operations (such as add and subtract)
>   in various scenarios such as task launch and terminate at multiple
>   modules such as master, allocator, sorter, etc.
> o Only allow DESTROY if there are no running or pending tasks using
>   the volume. However, if the volume is in a pending offer to one or
>   more frameworks, rescind that offer before processing the DESTROY.
> o To allow multiple tasks to be launched in the same ACCEPT call
>   using the same shared resource, we update the allocator and
>   sorter with additional copies of shared resources to reflect the
>   true shared count of allocated shared resources.
> 
> 
> Diffs
> -----
> 
>   src/master/allocator/mesos/hierarchical.hpp dd07ed221d2c1755d2478369641ffdc46ecc4471

>   src/master/allocator/mesos/hierarchical.cpp 9e5db2196c6a541dc1208ba8b9f13ef9a518bcc4

>   src/master/allocator/sorter/drf/sorter.hpp 09aa685f2bd7197385959d7d70d5411d0fd72f06

>   src/master/allocator/sorter/drf/sorter.cpp 1f2954c564eea5a3c649a5cc7181cb69329f9e35

>   src/master/http.cpp 525ef6cd7545d25f3ac89a6325ace6e6d768262a 
>   src/master/master.hpp c32c7e9d859ef73216354e2c03ecc07d0009b12f 
>   src/master/master.cpp b2a19a645528e8fc1fd48f5ac9929d38c9a76b49 
>   src/master/validation.hpp 3205ee1e699d0fb7ca16ced96a07be3a07cf8956 
>   src/master/validation.cpp 15e2ecc2de99d6bed522f45f855ba686bf19c008 
> 
> Diff: https://reviews.apache.org/r/45961/diff/
> 
> 
> Testing
> -------
> 
> Tests successful.
> 
> 
> Thanks,
> 
> Anindya Sinha
> 
>


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