mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gaston Kleiman <gas...@mesosphere.io>
Subject Re: Review Request 64932: Added example framework converting disk resources.
Date Thu, 04 Jan 2018 20:49:39 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/64932/#review194780
-----------------------------------------------------------



Should we add this new framework to `src/tests/examples_tests.cpp`?


src/examples/test_csi_user_framework.cpp
Lines 83-84 (patched)
<https://reviews.apache.org/r/64932/#comment273771>

    I see that this is also indented wrong in `test_http_framework.cpp`, but according to
the style guide, the proper indentation would e:
    
    ```
      HTTPScheduler(
          const FrameworkInfo& _framework,
          const string& _master)
    ```



src/examples/test_csi_user_framework.cpp
Lines 90-92 (patched)
<https://reviews.apache.org/r/64932/#comment273772>

    Ditto indentation.



src/examples/test_csi_user_framework.cpp
Lines 149 (patched)
<https://reviews.apache.org/r/64932/#comment273773>

    This framework never starts tasks, so should we crash or log something saying that we
don't expect to get task status updates?



src/examples/test_csi_user_framework.cpp
Lines 165-184 (patched)
<https://reviews.apache.org/r/64932/#comment273774>

    This framework will never launch a task, so this is probably not needed.
    
    Furthermore it should probably crash if it receivres an executor failure...



src/examples/test_csi_user_framework.cpp
Lines 278 (patched)
<https://reviews.apache.org/r/64932/#comment273786>

    I'd rephrase this:
    
    ```
    Check whether the given resources are reserved, ensuring that all RAW/MOUNT disk resources
offered by the resource provider are either reserved or unreserved.
    ```
    
    Wouldn't this break if a resource provider offers both reserved and unreserved disk resources?
I guess this will be usual once the agent's default disk resources (non-csi) are handled by
an SLRP.



src/examples/test_csi_user_framework.cpp
Lines 279 (patched)
<https://reviews.apache.org/r/64932/#comment273788>

    Why do you use an option here?
    
    Wouldn't `CHECK(!resources.isEmpty())` be clearer?



src/examples/test_csi_user_framework.cpp
Lines 302 (patched)
<https://reviews.apache.org/r/64932/#comment273789>

    Use `endl` instead of `'\n'`.



src/examples/test_csi_user_framework.cpp
Lines 318 (patched)
<https://reviews.apache.org/r/64932/#comment273790>

    Use `endl` instead of `'\n'`.



src/examples/test_csi_user_framework.cpp
Lines 331-332 (patched)
<https://reviews.apache.org/r/64932/#comment273792>

    I'd phrase it:
    
    If we didn't create an `ACCEPT` operation, create a `DELCINE` operation.



src/examples/test_csi_user_framework.cpp
Lines 334 (patched)
<https://reviews.apache.org/r/64932/#comment273793>

    Ditto `endl`.



src/examples/test_csi_user_framework.cpp
Lines 335 (patched)
<https://reviews.apache.org/r/64932/#comment273794>

    Remove this line, the if statement ensures that no accept call was created.



src/examples/test_csi_user_framework.cpp
Lines 380-386 (patched)
<https://reviews.apache.org/r/64932/#comment273795>

    This method doesn't seem to be used. `int main(...)` should probably call `Flags::setUsageMessage`
instead of this.



src/examples/test_csi_user_framework.cpp
Lines 399 (patched)
<https://reviews.apache.org/r/64932/#comment273796>

    Why don't we make this a `string` instead of an `Option<string>`?
    
    That way we could remove the: `if (flags.master.isNone())` block in the main function.



src/examples/test_csi_user_framework.cpp
Lines 444-447 (patched)
<https://reviews.apache.org/r/64932/#comment273797>

    Shouldn't this be a flag like in `no_executor_framework.cpp` and in `long_lived_framework.cpp`?


- Gaston Kleiman


On Jan. 3, 2018, 2:50 p.m., Benjamin Bannier wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/64932/
> -----------------------------------------------------------
> 
> (Updated Jan. 3, 2018, 2:50 p.m.)
> 
> 
> Review request for mesos, Greg Mann and Jie Yu.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This patch introduces an example HTTP framework which transforms
> 'RAW' disk resources from resource providers into 'MOUNT' volumes and
> subsequently unreserves them.
> 
> 
> Diffs
> -----
> 
>   src/Makefile.am 30cd4d426e797e4c8ee556d1bc3de99830a5fe41 
>   src/examples/CMakeLists.txt d4f1af4f072efdc68fa0b722f42b1d8aa1779b6e 
>   src/examples/test_csi_user_framework.cpp PRE-CREATION 
> 
> 
> Diff: https://reviews.apache.org/r/64932/diff/1/
> 
> 
> Testing
> -------
> 
> `make check`
> 
> 
> Thanks,
> 
> Benjamin Bannier
> 
>


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