> On Aug. 3, 2017, 10 a.m., Benjamin Bannier wrote:
> > src/resource_provider/daemon.hpp
> > Lines 44 (patched)
> > <https://reviews.apache.org/r/61275/diff/2/?file=1788101#file1788101line44>
> >
> >     I wonder if it would make more sense to keep passing a pid here like 
> > you did previously. With that we'd form the `URL` inside the daemon when 
> > creating the drivers.
> >     
> >     Since the use of http is a detail of how drivers currently work, 
> > passing a pid would prevent us from leaking that fact beyond the daemon. 
> > Also, a driver will take a `ContentType` which is specific to HTTP which 
> > makes sense together with an URL; an URL in isolation seems leaky for that 
> > reason as well to me. I believe a pid and some `slave::Flags` passed here 
> > would form a similarly intuitive package on a different level of 
> > abstraction.

`pid` is libprocess specific concept. The RP that uses this driver might not be 
using libprocess. As the other comment I have in the other review, this really 
should be an Endpoint detector if you want to make http driver general (that 
can be used for scheduler/executor drivers).

The v1 resource provider API is HTTP based API, so 'http' is not a detail, it's 
a protocol.


- Jie


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/61275/#review182080
-----------------------------------------------------------


On Aug. 2, 2017, 11:37 a.m., Jan Schlicht wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/61275/
> -----------------------------------------------------------
> 
> (Updated Aug. 2, 2017, 11:37 a.m.)
> 
> 
> Review request for mesos, Benjamin Bannier and Jie Yu.
> 
> 
> Bugs: MESOS-7469
>     https://issues.apache.org/jira/browse/MESOS-7469
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> The URL will be used by the resource provider driver to determine
> the endpoint of the resource provider manager API.
> 
> 
> Diffs
> -----
> 
>   include/mesos/v1/resource_provider.hpp 
> 88b606212ea57fee1c1ea522d2dc7f8124a9adef 
>   src/resource_provider/daemon.hpp 467a5d30510f0c8f7f042d989c615c97bbae52af 
>   src/resource_provider/daemon.cpp adcb60af8244840c56fbd2d846bdaeecb4f62282 
>   src/resource_provider/driver.cpp 6778ec9c863022446f141ee88f70eb563178ea05 
>   src/resource_provider/local.hpp 604c5d06b393349c352d0b698609ad6bfb16454a 
>   src/resource_provider/local.cpp a57c7c99c674f18036c36ec4b6df71947f700db8 
>   src/resource_provider/storage/provider.hpp 
> c6ea440feeb4cd79c59aa1f25851513e5d8dcc86 
>   src/resource_provider/storage/provider.cpp 
> 4c39312be5e4a6d783df3d385a66be6b3dcf8603 
>   src/slave/slave.cpp 7381530515f86faf4c3e8f82bcd9483f6cf0498b 
> 
> 
> Diff: https://reviews.apache.org/r/61275/diff/2/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Jan Schlicht
> 
>

Reply via email to