----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/50878/#review145047 -----------------------------------------------------------
Just FYI, https://reviews.apache.org/r/30430/ introduced the logic of `resource subtract` for `contains`. - Guangya Liu On 八月 7, 2016, 3:40 p.m., Guangya Liu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/50878/ > ----------------------------------------------------------- > > (Updated 八月 7, 2016, 3:40 p.m.) > > > Review request for mesos and Benjamin Mahler. > > > Repository: mesos > > > Description > ------- > > When check resources contains logic, there wil be logic of subtracting > the resources which is being checked. But this is not a must if the > resources is not a persistent volume as the non persistent volume > resources will be merged by default. > > This fix is adding persistent volume check before subtrat resources, > so that the non persistent volume resources will not be subtracted. > > > Diffs > ----- > > src/common/resources.cpp 2470c0280db7d48d9484c42bc2150e53e7ce6e1c > src/v1/resources.cpp 6c4e3b299e701d477947dd7427c31d2ae64c05ae > > Diff: https://reviews.apache.org/r/50878/diff/ > > > Testing > ------- > > make > make check > > The test result is that sorter benchmark improved 40%+ when `unallocated` > resources, the resources benchmark test improved 40%+ when checking contain > for `scalar` and `shared` resources. > > Sorter benchmark test before fix: > ``` > ./bin/mesos-tests.sh --benchmark > --gtest_filter="*Sorter_BENCHMARK_Test.FullSort/35" > [==========] Running 1 test from 1 test case. > [----------] Global test environment set-up. > [----------] 1 test from AgentAndClientCount/Sorter_BENCHMARK_Test > [ RUN ] AgentAndClientCount/Sorter_BENCHMARK_Test.FullSort/35 > Using 50000 agents and 1000 clients > Added 1000 clients in 24290us > Added 50000 agents in 1.483639secs > Added allocations for 50000 agents in 5.007551secs > Full sort of 1000 clients took 33151us > No-op sort of 1000 clients took 351us > Removed allocations for 50000 agents in 5.56734748333333mins > Removed 50000 agents in 2.050195secs > Removed 1000 clients in 19310us > [ OK ] AgentAndClientCount/Sorter_BENCHMARK_Test.FullSort/35 (342673 ms) > [----------] 1 test from AgentAndClientCount/Sorter_BENCHMARK_Test (342673 ms > total) > > [----------] Global test environment tear-down > [==========] 1 test from 1 test case ran. (342697 ms total) > [ PASSED ] 1 test. > ``` > > Sorter benchmark test after fix: > ``` > ./bin/mesos-tests.sh --benchmark > --gtest_filter="*Sorter_BENCHMARK_Test.FullSort/35" > [==========] Running 1 test from 1 test case. > [----------] Global test environment set-up. > [----------] 1 test from AgentAndClientCount/Sorter_BENCHMARK_Test > [ RUN ] AgentAndClientCount/Sorter_BENCHMARK_Test.FullSort/35 > Using 50000 agents and 1000 clients > Added 1000 clients in 20536us > Added 50000 agents in 1.592772secs > Added allocations for 50000 agents in 5.257316secs > Full sort of 1000 clients took 35049us > No-op sort of 1000 clients took 349us > Removed allocations for 50000 agents in 3.25236376666667mins > Removed 50000 agents in 1.835446secs > Removed 1000 clients in 19347us > [ OK ] AgentAndClientCount/Sorter_BENCHMARK_Test.FullSort/35 (203915 ms) > [----------] 1 test from AgentAndClientCount/Sorter_BENCHMARK_Test (203916 ms > total) > > [----------] Global test environment tear-down > [==========] 1 test from 1 test case ran. (203941 ms total) > [ PASSED ] 1 test. > > ``` > > Apply patch https://reviews.apache.org/r/50551/ > > Resources benchmark test before fix: > ``` > ./bin/mesos-tests.sh --benchmark > --gtest_filter="*Resources_BENCHMARK_Test.Contains/*” > [==========] Running 4 tests from 1 test case. > [----------] Global test environment set-up. > [----------] 4 tests from ResourcesOperators/Resources_BENCHMARK_Test > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/0 > Took 867642us to perform 50000 'initial.contains(r)' operations on cpus(*):1; > gpus(*):1; mem(*):128; disk(*):256 with initial resources cpus(*):1; > gpus(*):1; mem(*):128; disk(*):256 > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/0 (870 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/1 > Took 28.222642secs to perform 10 'initial.contains(r)' operations on cpus(0, > principal_0, {key_0: value_0}):1; gpus(... with initial resources cpus(0, > principal_0, {key_0: value_0}):1; gpus(... > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/1 (28280 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/2 > Took 480321us to perform 1000 'initial.contains(r)' operations on > ports(*):[1-2, 4-5, 7-8, 10-11, 13-14, 16-17, 1... with initial resources > ports(*):[30000-50000] > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/2 (481 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/3 > Took 1.133279secs to perform 50000 'initial.contains(r)' operations on > cpus(*):1; mem(*):128; disk(test)[persistentId:... with initial resources > cpus(*):1; mem(*):128; disk(test)[persistentId:... > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/3 (1134 ms) > [----------] 4 tests from ResourcesOperators/Resources_BENCHMARK_Test (30765 > ms total) > > [----------] Global test environment tear-down > [==========] 4 tests from 1 test case ran. (30791 ms total) > [ PASSED ] 4 tests. > ``` > > Resources benchmark test after fix: > ``` > ./bin/mesos-tests.sh --benchmark > --gtest_filter="*Resources_BENCHMARK_Test.Contains/*” > [==========] Running 4 tests from 1 test case. > [----------] Global test environment set-up. > [----------] 4 tests from ResourcesOperators/Resources_BENCHMARK_Test > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/0 > Took 578234us to perform 50000 'initial.contains(r)' operations on cpus(*):1; > gpus(*):1; mem(*):128; disk(*):256 with initial resources cpus(*):1; > gpus(*):1; mem(*):128; disk(*):256 > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/0 (578 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/1 > Took 30.010717secs to perform 10 'initial.contains(r)' operations on cpus(0, > principal_0, {key_0: value_0}):1; gpus(... with initial resources cpus(0, > principal_0, {key_0: value_0}):1; gpus(... > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/1 (30069 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/2 > Took 492528us to perform 1000 'initial.contains(r)' operations on > ports(*):[1-2, 4-5, 7-8, 10-11, 13-14, 16-17, 1... with initial resources > ports(*):[30000-50000] > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/2 (493 ms) > [ RUN ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/3 > Took 785835us to perform 50000 'initial.contains(r)' operations on cpus(*):1; > mem(*):128; disk(test)[persistentId:... with initial resources cpus(*):1; > mem(*):128; disk(test)[persistentId:... > [ OK ] ResourcesOperators/Resources_BENCHMARK_Test.Contains/3 (786 ms) > [----------] 4 tests from ResourcesOperators/Resources_BENCHMARK_Test (31926 > ms total) > > [----------] Global test environment tear-down > [==========] 4 tests from 1 test case ran. (31948 ms total) > [ PASSED ] 4 tests. > ``` > > > Thanks, > > Guangya Liu > >
