You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@mesos.apache.org by Vinod Kone <vi...@gmail.com> on 2015/08/18 20:28:12 UTC
Review Request 37588: Fixed master to drop scheduler HTTP calls during
recovery.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/37588/
-----------------------------------------------------------
Review request for mesos, Anand Mazumdar and Ben Mahler.
Bugs: MESOS-3290
https://issues.apache.org/jira/browse/MESOS-3290
Repository: mesos
Description
-------
Master returns an error when it is not the leader or is still doing recovery. This is same as what happens with PID frameworks.
Diffs
-----
src/master/http.cpp d016b3c055a06e7aa0649a36fd4b095c2938ba56
src/scheduler/scheduler.cpp 37f41debc394a773f33465dab1a89d7ef7264f64
Diff: https://reviews.apache.org/r/37588/diff/
Testing
-------
make check
Thanks,
Vinod Kone
Re: Review Request 37588: Fixed master to drop scheduler HTTP calls
during recovery.
Posted by Vinod Kone <vi...@gmail.com>.
> On Aug. 18, 2015, 6:49 p.m., Ben Mahler wrote:
> > It seems like Forbidden should be used for authorization issues, can we just have the non-leaders say they are not available? It seems to make sense, since they are not elected, they are not available.
changed master to send 503 when it's not elected.
> On Aug. 18, 2015, 6:49 p.m., Ben Mahler wrote:
> > src/master/http.cpp, lines 345-359
> > <https://reviews.apache.org/r/37588/diff/1/?file=1043280#file1043280line345>
> >
> > Would be nice to have some metrics for visibility into these, given we have dropped_messages for pid-based schedulers. TODO?
added a TODO.
- Vinod
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/37588/#review95758
-----------------------------------------------------------
On Aug. 18, 2015, 7:10 p.m., Vinod Kone wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/37588/
> -----------------------------------------------------------
>
> (Updated Aug. 18, 2015, 7:10 p.m.)
>
>
> Review request for mesos, Anand Mazumdar and Ben Mahler.
>
>
> Bugs: MESOS-3290
> https://issues.apache.org/jira/browse/MESOS-3290
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Master returns an error when it is not the leader or is still doing recovery. This is same as what happens with PID frameworks.
>
>
> Diffs
> -----
>
> src/master/http.cpp d016b3c055a06e7aa0649a36fd4b095c2938ba56
> src/scheduler/scheduler.cpp 37f41debc394a773f33465dab1a89d7ef7264f64
>
> Diff: https://reviews.apache.org/r/37588/diff/
>
>
> Testing
> -------
>
> make check
>
>
> Thanks,
>
> Vinod Kone
>
>
Re: Review Request 37588: Fixed master to drop scheduler HTTP calls
during recovery.
Posted by Ben Mahler <be...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/37588/#review95758
-----------------------------------------------------------
Ship it!
It seems like Forbidden should be used for authorization issues, can we just have the non-leaders say they are not available? It seems to make sense, since they are not elected, they are not available.
src/master/http.cpp (lines 345 - 359)
<https://reviews.apache.org/r/37588/#comment150868>
Would be nice to have some metrics for visibility into these, given we have dropped_messages for pid-based schedulers. TODO?
src/scheduler/scheduler.cpp (line 378)
<https://reviews.apache.org/r/37588/#comment150874>
It seems strange to me that forbidden is coming from non-leaders, forbidden seems like an authorization issue, no? Don't we want non-leaders to just say they are not available?
src/scheduler/scheduler.cpp (line 379)
<https://reviews.apache.org/r/37588/#comment150872>
Or the leadership changed since we sent this message..?
- Ben Mahler
On Aug. 18, 2015, 6:28 p.m., Vinod Kone wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/37588/
> -----------------------------------------------------------
>
> (Updated Aug. 18, 2015, 6:28 p.m.)
>
>
> Review request for mesos, Anand Mazumdar and Ben Mahler.
>
>
> Bugs: MESOS-3290
> https://issues.apache.org/jira/browse/MESOS-3290
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Master returns an error when it is not the leader or is still doing recovery. This is same as what happens with PID frameworks.
>
>
> Diffs
> -----
>
> src/master/http.cpp d016b3c055a06e7aa0649a36fd4b095c2938ba56
> src/scheduler/scheduler.cpp 37f41debc394a773f33465dab1a89d7ef7264f64
>
> Diff: https://reviews.apache.org/r/37588/diff/
>
>
> Testing
> -------
>
> make check
>
>
> Thanks,
>
> Vinod Kone
>
>
Re: Review Request 37588: Fixed master to drop scheduler HTTP calls
during recovery.
Posted by Vinod Kone <vi...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/37588/
-----------------------------------------------------------
(Updated Aug. 18, 2015, 7:10 p.m.)
Review request for mesos, Anand Mazumdar and Ben Mahler.
Changes
-------
removed the dependency and addressed comments. NNFR.
Bugs: MESOS-3290
https://issues.apache.org/jira/browse/MESOS-3290
Repository: mesos
Description
-------
Master returns an error when it is not the leader or is still doing recovery. This is same as what happens with PID frameworks.
Diffs (updated)
-----
src/master/http.cpp d016b3c055a06e7aa0649a36fd4b095c2938ba56
src/scheduler/scheduler.cpp 37f41debc394a773f33465dab1a89d7ef7264f64
Diff: https://reviews.apache.org/r/37588/diff/
Testing
-------
make check
Thanks,
Vinod Kone