> On Feb. 19, 2016, 1:02 a.m., Alexander Rukletsov wrote: > > src/tests/fetcher_cache_tests.cpp, lines 189-191 > > <https://reviews.apache.org/r/43367/diff/12/?file=1254386#file1254386line189> > > > > Will it be more clear to explicitly advance clock in those tests? I > > think keeping changes local, i.e. making them only there where they are > > necessary, serves educational purposes. People often read tests to > > understand how things work, hence it's valuable not to distract them with > > irrelevant actions, nor to hide meaningful steps. > > haosdent huang wrote: > Do you mean move > > ``` > Future<vector<Offer>> offers; > EXPECT_CALL(scheduler, resourceOffers(driver, _)) > .WillOnce(FutureArg<1>(&offers)) > .WillRepeatedly(DeclineOffers()); > > // The default timeout in AWAIT_READY is 15 seconds, > // so we use that amount here. > // TODO(bernd-mesos): Make this a symbolic constant in "gtest.hpp" > // that we can reference here. > offers.await(Seconds(15)); > if (!offers.isReady()) { > return Error("Failed to wait for resource offers: " + > (offers.isFailed() ? offers.failure() : "discarded")); > } > > CHECK_NE(0u, offers.get().size()); > const Offer offer = offers.get()[0]; > ``` > > out and use advance clock according to different test cases? > > Or pass a new param > ``` > Try<FetcherCacheTest::Task> FetcherCacheTest::launchTask( > const CommandInfo& commandInfo, > const size_t taskIndex, > const bool& advanceAllocation) > ```
This change applies to all these tests, so one global change instead of many local ones seems OK to me. Dropping the issue. - Bernd ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/43367/#review119838 ----------------------------------------------------------- On Feb. 18, 2016, 6:54 a.m., haosdent huang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/43367/ > ----------------------------------------------------------- > > (Updated Feb. 18, 2016, 6:54 a.m.) > > > Review request for mesos, Alexander Rukletsov, Benjamin Bannier, and Bernd > Mathiske. > > > Bugs: MESOS-4628 > https://issues.apache.org/jira/browse/MESOS-4628 > > > Repository: mesos > > > Description > ------- > > Speed up FetcherCache test cases by reduce allocation_interval. > > > Diffs > ----- > > src/tests/fetcher_cache_tests.cpp f9c48f5d938c2601cb8f826029d6969d676ab98e > > Diff: https://reviews.apache.org/r/43367/diff/ > > > Testing > ------- > > In this patch, we use 500ms instead of the default allocation_interval 1s. If > the test cases depends on getting offers from master multiple times, this > patch could reduce the offer waiting time and speed up them. > > Before > ``` > [ OK ] FetcherCacheTest.LocalUncached (943 ms) > [ OK ] FetcherCacheTest.LocalCached (1612 ms) > [ OK ] FetcherCacheTest.CachedFallback (906 ms) > [ OK ] FetcherCacheTest.LocalUncachedExtract (940 ms) > [ OK ] FetcherCacheTest.LocalCachedExtract (1719 ms) > [ OK ] FetcherCacheTest.SimpleEviction (3635 ms) > [ OK ] FetcherCacheTest.FallbackFromEviction (2518 ms) > [ OK ] FetcherCacheTest.RemoveLRUCacheEntries (3653 ms) > [ OK ] FetcherCacheHttpTest.HttpCachedSerialized (2632 ms) > [ OK ] FetcherCacheHttpTest.HttpCachedConcurrent (1227 ms) > [ OK ] FetcherCacheHttpTest.HttpMixed (1129 ms) > > ``` > > After > ``` > [ OK ] FetcherCacheTest.LocalUncached (979 ms) > [ OK ] FetcherCacheTest.LocalCached (1510 ms) > [ OK ] FetcherCacheTest.CachedFallback (853 ms) > [ OK ] FetcherCacheTest.LocalUncachedExtract (816 ms) > [ OK ] FetcherCacheTest.LocalCachedExtract (1715 ms) > [ OK ] FetcherCacheTest.SimpleEviction (2720 ms) > [ OK ] FetcherCacheTest.FallbackFromEviction (2119 ms) > [ OK ] FetcherCacheTest.RemoveLRUCacheEntries (2847 ms) > [ OK ] FetcherCacheHttpTest.HttpCachedSerialized (2233 ms) > [ OK ] FetcherCacheHttpTest.HttpCachedConcurrent (1211 ms) > [ OK ] FetcherCacheHttpTest.HttpMixed (1106 ms) > ``` > > > Thanks, > > haosdent huang > >
