Re: Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/ --- (Updated April 3, 2015, 11:47 p.m.) Review request for mesos and Ben Mahler. Bugs: MESOs-1127 https://issues.apache.org/jira/browse/MESOs-1127 Repository: mesos Description --- This is because subscription is done before schedulers can send any calls. Diffs - include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a include/mesos/scheduler/scheduler.proto 783a63ad1cc0edd86605d638046fb959cb6e97e8 src/examples/low_level_scheduler_libprocess.cpp 63d34eefb60d13fe2b82905c1cec9b762340e997 src/examples/low_level_scheduler_pthread.cpp 6d1f938660c02db75bfcbf7c8de0d941cff1920d src/master/master.cpp 618db68ee4163b06e479cf3413eda4b63c9c5a4b src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 Diff: https://reviews.apache.org/r/32507/diff/ Testing --- make check Thanks, Vinod Kone
Re: Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
> On March 31, 2015, 10:51 p.m., Ben Mahler wrote: > > Instead of having a separate /scheduler/events endpoint with a top level > > Subscribe protobuf, could we just have one /scheduler/call endpoint and > > have a SUBSCRIBE Call? > > > > I prefer this because: > > > > (1) There is a single endpoint / Call interface for the scheduler. > > > > (2) We don't have to have the "special" Subscribe top level protobuf. > > > > (3) When we move to HTTP/2, we don't have to change anything, we can just > > tell people that they are now free to use a *single* connection for all > > calls. For HTTP/1.1, we tell people that they must make the streaming > > Subscribe call on a standalone connection. > > > > Was the motivation for separating the Subscribe call out to a separate > > endpoint to try to make the connection management more obvious? People > > already have to think about connection management regardless of having > > /scheduler/events as a different path. > > > > Are there any major reasons I'm missing for not consolidating on one > > endpoint? Good point. Discarding this review in favor of putting Subscribe inside Call. - Vinod --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/#review78423 --- On March 31, 2015, 12:10 a.m., Vinod Kone wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/32507/ > --- > > (Updated March 31, 2015, 12:10 a.m.) > > > Review request for mesos and Ben Mahler. > > > Bugs: MESOs-1127 > https://issues.apache.org/jira/browse/MESOs-1127 > > > Repository: mesos > > > Description > --- > > This is because subscription is done before schedulers can send any calls. > > > Diffs > - > > include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a > include/mesos/scheduler/scheduler.proto > 783a63ad1cc0edd86605d638046fb959cb6e97e8 > src/examples/low_level_scheduler_libprocess.cpp > 63d34eefb60d13fe2b82905c1cec9b762340e997 > src/examples/low_level_scheduler_pthread.cpp > 6d1f938660c02db75bfcbf7c8de0d941cff1920d > src/master/master.cpp 618db68ee4163b06e479cf3413eda4b63c9c5a4b > src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 > src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a > src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 > > Diff: https://reviews.apache.org/r/32507/diff/ > > > Testing > --- > > make check > > > Thanks, > > Vinod Kone > >
Re: Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/#review78423 --- Instead of having a separate /scheduler/events endpoint with a top level Subscribe protobuf, could we just have one /scheduler/call endpoint and have a SUBSCRIBE Call? I prefer this because: (1) There is a single endpoint / Call interface for the scheduler. (2) We don't have to have the "special" Subscribe top level protobuf. (3) When we move to HTTP/2, we don't have to change anything, we can just tell people that they are now free to use a *single* connection for all calls. For HTTP/1.1, we tell people that they must make the streaming Subscribe call on a standalone connection. Was the motivation for separating the Subscribe call out to a separate endpoint to try to make the connection management more obvious? People already have to think about connection management regardless of having /scheduler/events as a different path. Are there any major reasons I'm missing for not consolidating on one endpoint? - Ben Mahler On March 31, 2015, 12:10 a.m., Vinod Kone wrote: > > --- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/32507/ > --- > > (Updated March 31, 2015, 12:10 a.m.) > > > Review request for mesos and Ben Mahler. > > > Bugs: MESOs-1127 > https://issues.apache.org/jira/browse/MESOs-1127 > > > Repository: mesos > > > Description > --- > > This is because subscription is done before schedulers can send any calls. > > > Diffs > - > > include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a > include/mesos/scheduler/scheduler.proto > 783a63ad1cc0edd86605d638046fb959cb6e97e8 > src/examples/low_level_scheduler_libprocess.cpp > 63d34eefb60d13fe2b82905c1cec9b762340e997 > src/examples/low_level_scheduler_pthread.cpp > 6d1f938660c02db75bfcbf7c8de0d941cff1920d > src/master/master.cpp 618db68ee4163b06e479cf3413eda4b63c9c5a4b > src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 > src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a > src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 > > Diff: https://reviews.apache.org/r/32507/diff/ > > > Testing > --- > > make check > > > Thanks, > > Vinod Kone > >
Re: Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/ --- (Updated March 31, 2015, 12:10 a.m.) Review request for mesos and Ben Mahler. Changes --- rebased. Bugs: MESOs-1127 https://issues.apache.org/jira/browse/MESOs-1127 Repository: mesos Description --- This is because subscription is done before schedulers can send any calls. Diffs (updated) - include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a include/mesos/scheduler/scheduler.proto 783a63ad1cc0edd86605d638046fb959cb6e97e8 src/examples/low_level_scheduler_libprocess.cpp 63d34eefb60d13fe2b82905c1cec9b762340e997 src/examples/low_level_scheduler_pthread.cpp 6d1f938660c02db75bfcbf7c8de0d941cff1920d src/master/master.cpp 618db68ee4163b06e479cf3413eda4b63c9c5a4b src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 Diff: https://reviews.apache.org/r/32507/diff/ Testing --- make check Thanks, Vinod Kone
Re: Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/ --- (Updated March 27, 2015, 12:05 a.m.) Review request for mesos and Ben Mahler. Changes --- rebase. Bugs: MESOs-1127 https://issues.apache.org/jira/browse/MESOs-1127 Repository: mesos Description --- This is because subscription is done before schedulers can send any calls. Diffs (updated) - include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a include/mesos/scheduler/scheduler.proto 783a63ad1cc0edd86605d638046fb959cb6e97e8 src/examples/low_level_scheduler_libprocess.cpp 63d34eefb60d13fe2b82905c1cec9b762340e997 src/examples/low_level_scheduler_pthread.cpp 6d1f938660c02db75bfcbf7c8de0d941cff1920d src/master/master.cpp dccd7c635da4b7031cd109bd84e7f17b31777ef1 src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 Diff: https://reviews.apache.org/r/32507/diff/ Testing --- make check Thanks, Vinod Kone
Review Request 32507: Moved REGISTER and REREGISTER out of scheduler Calls.
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32507/ --- Review request for mesos and Ben Mahler. Bugs: MESOs-1127 https://issues.apache.org/jira/browse/MESOs-1127 Repository: mesos Description --- This is because subscription is done before schedulers can send any calls. Diffs - include/mesos/scheduler.hpp efee2cb5fc9f24e84294ed7e05a25cf8c81c2f1a include/mesos/scheduler/scheduler.proto 783a63ad1cc0edd86605d638046fb959cb6e97e8 src/examples/low_level_scheduler_libprocess.cpp 63d34eefb60d13fe2b82905c1cec9b762340e997 src/examples/low_level_scheduler_pthread.cpp 6d1f938660c02db75bfcbf7c8de0d941cff1920d src/master/master.cpp dccd7c635da4b7031cd109bd84e7f17b31777ef1 src/sched/sched.cpp 66fd2b3146568900356cc48e0f17c6720665ae80 src/scheduler/scheduler.cpp 584b042e32865fdf875bf41ebcfb7f9c327d882a src/tests/scheduler_tests.cpp 4a89a7a88b50bb8c254f5076661ce07ac9fc7657 Diff: https://reviews.apache.org/r/32507/diff/ Testing --- make check Thanks, Vinod Kone
