----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/71080/#review216836 -----------------------------------------------------------
LGTM as a work around, which we won't commit upstream. We reached a decision in https://mesos.slack.com/archives/C8PDVCDE3/p1563816071006900 to not merge it upstream as we couldn't reach consensus on how the flag / semantics should be maintained. include/mesos/allocator/allocator.hpp Lines 66 (patched) <https://reviews.apache.org/r/71080/#comment304084> `maxCompletedFrameworks` is used for the metrics to keep for completed frameworks in the allocator. `maxCompletedFramework_ids` (we should name it `maxCompletedFrameworkIDs` if it were useful) is not used and we don't need to add this field. src/master/master.hpp Lines 2210 (patched) <https://reviews.apache.org/r/71080/#comment304085> camelCasing to be consistent. src/master/master.hpp Lines 2215 (patched) <https://reviews.apache.org/r/71080/#comment304086> I would use a `bool` (1 byte) as the value. And a comment that the value is not meaningful and we are just using the map to simulate a "bounded hashset". - Jiang Yan Xu On July 18, 2019, 8:10 p.m., Xudong Ni wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/71080/ > ----------------------------------------------------------- > > (Updated July 18, 2019, 8:10 p.m.) > > > Review request for mesos, James Peach and Jiang Yan Xu. > > > Bugs: MESOS-8636 > https://issues.apache.org/jira/browse/MESOS-8636 > > > Repository: mesos > > > Description > ------- > > It should be separately from that for webUI and endpoints. > Currently the master stores the history of completed frameworks in > a map with the full historical data of the framework, it is > prohibitively expensive to keep a long history; In order to reject > frameworks from reregistering if they have previously marked as > completed, we only need to persist the framework IDs and are able > to keep long history. > > > Diffs > ----- > > docs/configuration/master.md c56ac8510ea968f9587e23e81ed310caa968ee9e > docs/operator-http-api.md 2d4a9b66e20cf19eceec718b7de3d812ab285772 > include/mesos/allocator/allocator.hpp > 2bab53ab5fb25931a724c20a039e1301983ba574 > src/master/constants.hpp 26afa356b7844b4ec6c2caeef33bd39c51148d5f > src/master/flags.hpp 78623d68bf428cd3f52684303d98a525d42eb878 > src/master/flags.cpp 74f4daadd48e8e691be43759b88dc8b3c2df489a > src/master/http.cpp 765d5052ab6a8d731113f8922d20fb280b843003 > src/master/master.hpp ffa7423ba533725f7c1123d9aa507b1348e7f281 > src/master/master.cpp f1ca637b4cb0382caec53b5a81f6a4eb46f4dd2d > > > Diff: https://reviews.apache.org/r/71080/diff/7/ > > > Testing > ------- > > make check > > sudo GLOG_v=1 ./bin/mesos-tests.sh --verbose > --gtest_filter=MasterTest.MaxCompletedFrameworksFlag --gtest_break_on_failure > --gtest_repeat=1000 > [ OK ] MasterTest.MaxCompletedFrameworksFlag (230 ms) > [----------] 1 test from MasterTest (235 ms total) > > [----------] Global test environment tear-down > [==========] 1 test from 1 test case ran. (281 ms total) > [ PASSED ] 1 test. > > > Thanks, > > Xudong Ni > >