You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@mesos.apache.org by Anand Mazumdar <ma...@gmail.com> on 2016/04/13 00:26:12 UTC
Review Request 46115: Added AuthN for HTTP based frameworks.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/
-----------------------------------------------------------
Review request for mesos, Greg Mann and Vinod Kone.
Bugs: MESOS-3923
https://issues.apache.org/jira/browse/MESOS-3923
Repository: mesos
Description
-------
This change adds AuthN support for HTTP based frameworks.
We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
we ensure that `FrameworkInfo.principal` (if present) equals
the authenticated principal. We also allow a framework to
omit specifying the `FrameworkInfo.principal` in case
it is not interested in authorization or if it is disabled.
We do log a warning for this cases similar to what driver
based frameworks already do.
Diffs
-----
src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
Diff: https://reviews.apache.org/r/46115/diff/
Testing
-------
make check. Enabled AuthN for all the tests later in the chain.
Thanks,
Anand Mazumdar
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Vinod Kone <vi...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/#review129051
-----------------------------------------------------------
Fix it, then Ship it!
src/master/http.cpp (lines 437 - 439)
<https://reviews.apache.org/r/46115/#comment192509>
Update this warning message per my comment in the other review.
- Vinod Kone
On April 14, 2016, 5:46 p.m., Anand Mazumdar wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/46115/
> -----------------------------------------------------------
>
> (Updated April 14, 2016, 5:46 p.m.)
>
>
> Review request for mesos, Greg Mann and Vinod Kone.
>
>
> Bugs: MESOS-3923
> https://issues.apache.org/jira/browse/MESOS-3923
>
>
> Repository: mesos
>
>
> Description
> -------
>
> This change adds AuthN support for HTTP based frameworks.
>
> We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
> we ensure that `FrameworkInfo.principal` (if present) equals
> the authenticated principal. We also allow a framework to
> omit specifying the `FrameworkInfo.principal` in case
> it is not interested in authorization or if it is disabled.
> We do log a warning for this cases similar to what driver
> based frameworks already do.
>
>
> Diffs
> -----
>
> src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
> src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
> src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
> src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
> src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
>
> Diff: https://reviews.apache.org/r/46115/diff/
>
>
> Testing
> -------
>
> make check. Enabled AuthN for all the tests later in the chain.
>
>
> Thanks,
>
> Anand Mazumdar
>
>
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Anand Mazumdar <ma...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/
-----------------------------------------------------------
(Updated April 15, 2016, 6:07 p.m.)
Review request for mesos, Greg Mann and Vinod Kone.
Changes
-------
Made the log message consistent with driver based framework
Bugs: MESOS-3923
https://issues.apache.org/jira/browse/MESOS-3923
Repository: mesos
Description
-------
This change adds AuthN support for HTTP based frameworks.
We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
we ensure that `FrameworkInfo.principal` (if present) equals
the authenticated principal. We also allow a framework to
omit specifying the `FrameworkInfo.principal` in case
it is not interested in authorization or if it is disabled.
We do log a warning for this cases similar to what driver
based frameworks already do.
Diffs (updated)
-----
src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
Diff: https://reviews.apache.org/r/46115/diff/
Testing
-------
make check. Enabled AuthN for all the tests later in the chain.
Thanks,
Anand Mazumdar
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Anand Mazumdar <ma...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/
-----------------------------------------------------------
(Updated April 14, 2016, 5:46 p.m.)
Review request for mesos, Greg Mann and Vinod Kone.
Changes
-------
Review comments from Greg
Bugs: MESOS-3923
https://issues.apache.org/jira/browse/MESOS-3923
Repository: mesos
Description
-------
This change adds AuthN support for HTTP based frameworks.
We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
we ensure that `FrameworkInfo.principal` (if present) equals
the authenticated principal. We also allow a framework to
omit specifying the `FrameworkInfo.principal` in case
it is not interested in authorization or if it is disabled.
We do log a warning for this cases similar to what driver
based frameworks already do.
Diffs (updated)
-----
src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
Diff: https://reviews.apache.org/r/46115/diff/
Testing
-------
make check. Enabled AuthN for all the tests later in the chain.
Thanks,
Anand Mazumdar
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Anand Mazumdar <ma...@gmail.com>.
> On April 14, 2016, 5:10 p.m., Greg Mann wrote:
> > src/master/validation.hpp, line 47
> > <https://reviews.apache.org/r/46115/diff/2/?file=1344573#file1344573line47>
> >
> > Do we really want to have a default value for an argument to a validation function? It makes me a bit uneasy; if the caller wants a `Call` message to be validated, shouldn't we require them to provide the exact data to validated, without making assumptions for them?
Unfortunately, this method is also used by the validation function in `receive` of master: https://github.com/apache/mesos/blob/master/src/master/master.cpp#L1952 by driver based frameworks.
> On April 14, 2016, 5:10 p.m., Greg Mann wrote:
> > src/master/validation.cpp, lines 79-81
> > <https://reviews.apache.org/r/46115/diff/2/?file=1344574#file1344574line79>
> >
> > What about: `if (principal.isSome() && frameworkInfo.has_principal() && principal != frameworkInfo.principal())`?
My bad, the previous version had two `if` statements. But, I ended up moving one but forgot to clean this up.
- Anand
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/#review128914
-----------------------------------------------------------
On April 14, 2016, 5:46 p.m., Anand Mazumdar wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/46115/
> -----------------------------------------------------------
>
> (Updated April 14, 2016, 5:46 p.m.)
>
>
> Review request for mesos, Greg Mann and Vinod Kone.
>
>
> Bugs: MESOS-3923
> https://issues.apache.org/jira/browse/MESOS-3923
>
>
> Repository: mesos
>
>
> Description
> -------
>
> This change adds AuthN support for HTTP based frameworks.
>
> We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
> we ensure that `FrameworkInfo.principal` (if present) equals
> the authenticated principal. We also allow a framework to
> omit specifying the `FrameworkInfo.principal` in case
> it is not interested in authorization or if it is disabled.
> We do log a warning for this cases similar to what driver
> based frameworks already do.
>
>
> Diffs
> -----
>
> src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
> src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
> src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
> src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
> src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
>
> Diff: https://reviews.apache.org/r/46115/diff/
>
>
> Testing
> -------
>
> make check. Enabled AuthN for all the tests later in the chain.
>
>
> Thanks,
>
> Anand Mazumdar
>
>
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Greg Mann <gr...@mesosphere.io>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/#review128914
-----------------------------------------------------------
Looks good! A couple small things below.
src/master/http.cpp (lines 431 - 436)
<https://reviews.apache.org/r/46115/#comment192355>
What about instead:
`if (principal.isSome() && !call.subscribe().framework_info().has_principal())`?
src/master/validation.hpp (line 47)
<https://reviews.apache.org/r/46115/#comment192362>
Do we really want to have a default value for an argument to a validation function? It makes me a bit uneasy; if the caller wants a `Call` message to be validated, shouldn't we require them to provide the exact data to validated, without making assumptions for them?
src/master/validation.cpp (lines 79 - 81)
<https://reviews.apache.org/r/46115/#comment192358>
What about: `if (principal.isSome() && frameworkInfo.has_principal() && principal != frameworkInfo.principal())`?
- Greg Mann
On April 14, 2016, 3:35 p.m., Anand Mazumdar wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/46115/
> -----------------------------------------------------------
>
> (Updated April 14, 2016, 3:35 p.m.)
>
>
> Review request for mesos, Greg Mann and Vinod Kone.
>
>
> Bugs: MESOS-3923
> https://issues.apache.org/jira/browse/MESOS-3923
>
>
> Repository: mesos
>
>
> Description
> -------
>
> This change adds AuthN support for HTTP based frameworks.
>
> We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
> we ensure that `FrameworkInfo.principal` (if present) equals
> the authenticated principal. We also allow a framework to
> omit specifying the `FrameworkInfo.principal` in case
> it is not interested in authorization or if it is disabled.
> We do log a warning for this cases similar to what driver
> based frameworks already do.
>
>
> Diffs
> -----
>
> src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
> src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
> src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
> src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
> src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
>
> Diff: https://reviews.apache.org/r/46115/diff/
>
>
> Testing
> -------
>
> make check. Enabled AuthN for all the tests later in the chain.
>
>
> Thanks,
>
> Anand Mazumdar
>
>
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Anand Mazumdar <ma...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/
-----------------------------------------------------------
(Updated April 14, 2016, 3:35 p.m.)
Review request for mesos, Greg Mann and Vinod Kone.
Changes
-------
Review comments from Vinod
Bugs: MESOS-3923
https://issues.apache.org/jira/browse/MESOS-3923
Repository: mesos
Description
-------
This change adds AuthN support for HTTP based frameworks.
We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
we ensure that `FrameworkInfo.principal` (if present) equals
the authenticated principal. We also allow a framework to
omit specifying the `FrameworkInfo.principal` in case
it is not interested in authorization or if it is disabled.
We do log a warning for this cases similar to what driver
based frameworks already do.
Diffs (updated)
-----
src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
Diff: https://reviews.apache.org/r/46115/diff/
Testing
-------
make check. Enabled AuthN for all the tests later in the chain.
Thanks,
Anand Mazumdar
Re: Review Request 46115: Added AuthN for HTTP based frameworks.
Posted by Vinod Kone <vi...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/46115/#review128731
-----------------------------------------------------------
src/master/master.cpp (line 850)
<https://reviews.apache.org/r/46115/#comment192197>
As discussed lets have a different REALM for http frameworks.
src/master/validation.cpp (lines 80 - 81)
<https://reviews.apache.org/r/46115/#comment192226>
After master sets FrameworkInfo.principal to the authenticating principal, please update this comment.
- Vinod Kone
On April 12, 2016, 10:26 p.m., Anand Mazumdar wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/46115/
> -----------------------------------------------------------
>
> (Updated April 12, 2016, 10:26 p.m.)
>
>
> Review request for mesos, Greg Mann and Vinod Kone.
>
>
> Bugs: MESOS-3923
> https://issues.apache.org/jira/browse/MESOS-3923
>
>
> Repository: mesos
>
>
> Description
> -------
>
> This change adds AuthN support for HTTP based frameworks.
>
> We allow AuthZ without AuthN for `/scheduler` endpoint. Also,
> we ensure that `FrameworkInfo.principal` (if present) equals
> the authenticated principal. We also allow a framework to
> omit specifying the `FrameworkInfo.principal` in case
> it is not interested in authorization or if it is disabled.
> We do log a warning for this cases similar to what driver
> based frameworks already do.
>
>
> Diffs
> -----
>
> src/master/http.cpp b8a83b58b60416f61610cad16fc6f70028a5ee10
> src/master/master.hpp 1f480f03900a0dc2996c2ed3a9534dfa8036940f
> src/master/master.cpp 781402c04fded159183e1ca28894e48355200f0c
> src/master/validation.hpp d1f2323172cbf3bb052942a119b8531f9ccad48d
> src/master/validation.cpp 13423436a4e6361fde6fa75133eebf5c02c8381f
>
> Diff: https://reviews.apache.org/r/46115/diff/
>
>
> Testing
> -------
>
> make check. Enabled AuthN for all the tests later in the chain.
>
>
> Thanks,
>
> Anand Mazumdar
>
>