You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@mesos.apache.org by Benjamin Mahler <bm...@apache.org> on 2019/01/15 19:27:11 UTC
Re: Review Request 69588: Removed outdated authorization logic for
offer operations.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/69588/#review212027
-----------------------------------------------------------
src/master/master.cpp
Lines 3604-3605 (original), 3604-3605 (patched)
<https://reviews.apache.org/r/69588/#comment297618>
Isn't the empty case valid? Also, use `Resources::reservationRole`?
src/master/master.cpp
Line 3662 (original), 3664 (patched)
<https://reviews.apache.org/r/69588/#comment297617>
The `reservations.empty()` case is actually an invalid call right? There's nothing to unreserve.
Should we be CHECKing that here or is that validated after this point today?
src/master/master.cpp
Lines 3723-3724 (original), 3723-3724 (patched)
<https://reviews.apache.org/r/69588/#comment297619>
Is the empty case valid?
Also, use `Resources::reservationRole`?
src/master/master.cpp
Line 3818 (original), 3826-3829 (patched)
<https://reviews.apache.org/r/69588/#comment297620>
Can you use `Resources::reservationRole` for this? I guess `*` is invalid for persistent volumes but we don't care here?
src/master/master.cpp
Lines 3877-3880 (original), 3892-3895 (patched)
<https://reviews.apache.org/r/69588/#comment297621>
ditto `Resources::reservationRole` and `*` being invalid, but I guess we don't care here?
src/master/master.cpp
Lines 3940-3943 (original), 3959-3962 (patched)
<https://reviews.apache.org/r/69588/#comment297622>
Ditto `Resources::reservationRole` and `*` being invalid (but I guess we don't care here?)
- Benjamin Mahler
On Dec. 19, 2018, 11:20 p.m., Chun-Hung Hsiao wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/69588/
> -----------------------------------------------------------
>
> (Updated Dec. 19, 2018, 11:20 p.m.)
>
>
> Review request for mesos, Benjamin Bannier, Benjamin Mahler, Michael Park, and Till Toenshoff.
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Since the master now upgrades resources to the post-reservation-refinement
> format before authorization (see MESOS-7735), the authorization logic in the
> master becomes outdated. This patch cleans it up.
>
>
> Diffs
> -----
>
> src/master/master.cpp b4faf2b077a0288ba36195b7a21402932489d316
>
>
> Diff: https://reviews.apache.org/r/69588/diff/2/
>
>
> Testing
> -------
>
> sudo make check
>
>
> Thanks,
>
> Chun-Hung Hsiao
>
>
Re: Review Request 69588: Removed outdated authorization logic for
offer operations.
Posted by Chun-Hung Hsiao <ch...@apache.org>.
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Line 3818 (original), 3826-3829 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3839>
> >
> > Can you use `Resources::reservationRole` for this? I guess `*` is invalid for persistent volumes but we don't care here?
Right. `*` is invalid for persistent volumes. But we used to still send out a authorization request before validating the operation. Given that this is an experimental feature, should we ignore the resource here instead?
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Lines 3877-3880 (original), 3892-3895 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3905>
> >
> > ditto `Resources::reservationRole` and `*` being invalid, but I guess we don't care here?
We do care about this case, as `CREATE_DISK` can be applied to an unreserved `RAW` disk.
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Lines 3940-3943 (original), 3959-3962 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3972>
> >
> > Ditto `Resources::reservationRole` and `*` being invalid (but I guess we don't care here?)
See above.
- Chun-Hung
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/69588/#review212027
-----------------------------------------------------------
On Dec. 19, 2018, 11:20 p.m., Chun-Hung Hsiao wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/69588/
> -----------------------------------------------------------
>
> (Updated Dec. 19, 2018, 11:20 p.m.)
>
>
> Review request for mesos, Benjamin Bannier, Benjamin Mahler, Michael Park, and Till Toenshoff.
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Since the master now upgrades resources to the post-reservation-refinement
> format before authorization (see MESOS-7735), the authorization logic in the
> master becomes outdated. This patch cleans it up.
>
>
> Diffs
> -----
>
> src/master/master.cpp b4faf2b077a0288ba36195b7a21402932489d316
>
>
> Diff: https://reviews.apache.org/r/69588/diff/2/
>
>
> Testing
> -------
>
> sudo make check
>
>
> Thanks,
>
> Chun-Hung Hsiao
>
>
Re: Review Request 69588: Removed outdated authorization logic for
offer operations.
Posted by Chun-Hung Hsiao <ch...@apache.org>.
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Lines 3604-3605 (original), 3604-3605 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3604>
> >
> > Isn't the empty case valid? Also, use `Resources::reservationRole`?
>
> Benjamin Bannier wrote:
> With that helper we can probably even get rid of or reduce above comment block and assertions, but should `CHECK(Resources::isReserved(resource))` since `Resources::reservationRole` is only defined for reserved resources. Still need to think about unreserved resources/empty case.
>
> Here and below.
The reason I explicitly set the role to `*` is for backward compatibility.
Since now the resource is always in the post-reservation-refinement format, the empty case indicates that the resource is unreserved, and the old behavior is to look into `resource.role()` which would return the default value `*`.
But, let me use the helpers to make it more readable:
```
const string role = Resources::isReserved(resources) ? Resources::reservationRole(resource) : "*";
```
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Line 3662 (original), 3664 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3670>
> >
> > The `reservations.empty()` case is actually an invalid call right? There's nothing to unreserve.
> >
> > Should we be CHECKing that here or is that validated after this point today?
Unfortunately, we validate the operation in `Master::_accept`, *after* getting authorizations: https://github.com/apache/mesos/blob/2c658770e7e273672d64fe6c7deaed05828e9a15/src/master/master.cpp#L4925,
so we have to deal with this case. I'm just setting it to `*` for consistency and backward compatibility.
Also see https://github.com/apache/mesos/blob/2c658770e7e273672d64fe6c7deaed05828e9a15/src/master/master.cpp#L3669.
Let me add the comments related to validation back.
> On Jan. 15, 2019, 7:27 p.m., Benjamin Mahler wrote:
> > src/master/master.cpp
> > Lines 3723-3724 (original), 3723-3724 (patched)
> > <https://reviews.apache.org/r/69588/diff/2/?file=2115120#file2115120line3734>
> >
> > Is the empty case valid?
> >
> > Also, use `Resources::reservationRole`?
See above.
- Chun-Hung
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/69588/#review212027
-----------------------------------------------------------
On Dec. 19, 2018, 11:20 p.m., Chun-Hung Hsiao wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/69588/
> -----------------------------------------------------------
>
> (Updated Dec. 19, 2018, 11:20 p.m.)
>
>
> Review request for mesos, Benjamin Bannier, Benjamin Mahler, Michael Park, and Till Toenshoff.
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Since the master now upgrades resources to the post-reservation-refinement
> format before authorization (see MESOS-7735), the authorization logic in the
> master becomes outdated. This patch cleans it up.
>
>
> Diffs
> -----
>
> src/master/master.cpp b4faf2b077a0288ba36195b7a21402932489d316
>
>
> Diff: https://reviews.apache.org/r/69588/diff/2/
>
>
> Testing
> -------
>
> sudo make check
>
>
> Thanks,
>
> Chun-Hung Hsiao
>
>