----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/53691/#review160153 -----------------------------------------------------------
src/tests/master_quota_tests.cpp (lines 1721 - 1723) <https://reviews.apache.org/r/53691/#comment231261> We should probably rephrase this; see my comment below. src/tests/master_quota_tests.cpp (lines 1837 - 1839) <https://reviews.apache.org/r/53691/#comment231259> Yeah, current behaviour indeed prevents satisfied quota'ed frameworks from receiving non-revocable resources beyond their quota. It probably makes sense to mention this assumption in the test comment. You don't have to trigger an allocation though, because adding an agent triggers one. I think it's also fine to leave a note here that we expect no allocations here. We don't really have to check because if an offer arrives here the next expectation will fail. What I originially had in mind was some kind of test that quota is actually helpful when agents start dying in the cluster. We can simulate this situation as well, however without quota enforcement (read: preemption and revocation) this test does not seem to have a lot of sense. I suggest to punt on my original idea for now and just go with your proposal. src/tests/master_quota_tests.cpp (lines 1843 - 1846) <https://reviews.apache.org/r/53691/#comment231251> Please keep formatting consistent throughout the file. src/tests/master_quota_tests.cpp (lines 1862 - 1864) <https://reviews.apache.org/r/53691/#comment231260> The expectation should be set before the event is triggered, otherwise there is a race: the offer will be processed and ignored and hence `AWAIT_READY(offers2);` will never be triggered. - Alexander Rukletsov On Dec. 27, 2016, 11:24 a.m., Zhitao Li wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/53691/ > ----------------------------------------------------------- > > (Updated Dec. 27, 2016, 11:24 a.m.) > > > Review request for mesos, Alexander Rukletsov and Xiaojian Huang. > > > Bugs: MESOS-3982 > https://issues.apache.org/jira/browse/MESOS-3982 > > > Repository: mesos > > > Description > ------- > > This implements some tests in previous todos for > quota functionality. > > The one for: > > `Role quota is below its allocation (InverseOffer generation).` > > does not seem be implementable right now, since > hierarchical allocator does not send InverseOffer for > quota yet. > > > Diffs > ----- > > src/tests/master_quota_tests.cpp 48be7406181646c8cc1d169b82a4a4ca71cdf03b > > Diff: https://reviews.apache.org/r/53691/diff/ > > > Testing > ------- > > > Thanks, > > Zhitao Li > >
