----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/39331/#review105962 -----------------------------------------------------------
src/slave/containerizer/mesos/provisioner/docker/store.cpp (line 136) <https://reviews.apache.org/r/39331/#comment164658> Actually we don't need to check exists before mkdir, so you can remove this check. src/slave/containerizer/mesos/provisioner/docker/store.cpp (line 240) <https://reviews.apache.org/r/39331/#comment164660> This is not right, you need to put the removing from the pulling hashmap in the onAny block in line 230, because this is block is not called if no one requests the same image in the same time, and will leave it in the hashmap. - Timothy Chen On Nov. 9, 2015, 7:26 p.m., Gilbert Song wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/39331/ > ----------------------------------------------------------- > > (Updated Nov. 9, 2015, 7:26 p.m.) > > > Review request for mesos, Anand Mazumdar, Jie Yu, Jojy Varghese, and Timothy > Chen. > > > Bugs: MESOS-3736 > https://issues.apache.org/jira/browse/MESOS-3736 > > > Repository: mesos > > > Description > ------- > > Support docker local store pull image simultaneously > > > Diffs > ----- > > src/slave/containerizer/mesos/provisioner/docker/store.hpp > 95e46b9914c018b3e2472f98a54bc33ff9a46e17 > src/slave/containerizer/mesos/provisioner/docker/store.cpp > bb02d650e16d45fcf337a7954f7a26143fb2c69f > src/tests/containerizer/provisioner_docker_tests.cpp > fe6a90fe32364eec8ef923a000db19183603c338 > > Diff: https://reviews.apache.org/r/39331/diff/ > > > Testing > ------- > > make check (ubuntu14.04 + clang-3.6) > > *This is not ready to be merged. > *Still considering two question: > 1. Handling simultaneous failure. If the first request is called and is > written into the hashmap. All the other requests will be waiting for the > future of the first request. But because its return type is > 'Future<vector<string>>', if its future status is 'FAILED/DISCARDED', the > other requests will be waiting forever. > Solved by logic check: if it is the first call to get() Image_A, promise > associate with metadateManager->get(). If not, check whether that promised > future failed/discarded. If yes, over write to the hashmap. > > 2. The current hashmap uses 'stringify(image::name)' as key, but it may not > be unique because there is chance that layer_ids can be changed. > One solution is to have 'stringify(image)' as key. > > > Thanks, > > Gilbert Song > >
