----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/8128/#review13591 -----------------------------------------------------------
src/master/master.cpp <https://reviews.apache.org/r/8128/#comment29164> please include the hostname << " (" << slave->info.hostname() << ")" src/master/master.cpp <https://reviews.apache.org/r/8128/#comment29165> Not entirely yours, but would you mind re-organizing the code inside this function? As is, its hard to reason about the flow. How about.. Framework* framework = getFramework(frameworkId); if (framework != NULL) { // Send ExitedExecutorMessage to framework // framework->removeExecutor(slaveId, executorId); } else { LOG(WARNING) << } Slave* slave = getSlave(slaveId); if (slave == NULL) { LOG(WARNING) << return; } if (slave->hasExecutor()) { // Blah } else { LOG(WARNING) << } - Vinod Kone On Nov. 19, 2012, 10:09 p.m., Charles Reiss wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/8128/ > ----------------------------------------------------------- > > (Updated Nov. 19, 2012, 10:09 p.m.) > > > Review request for mesos and Benjamin Hindman. > > > Description > ------- > > Forward ExitedExecutorMessages from the master to framework schedulers, and > call Scheduler::executorLost when they are received. > > > This addresses bug MESOS-313. > https://issues.apache.org/jira/browse/MESOS-313 > > > Diffs > ----- > > src/master/master.cpp 87fad0e > src/sched/sched.cpp a1247f9 > src/tests/gc_tests.cpp 6b3ba02 > src/tests/master_tests.cpp 948ab5d > > Diff: https://reviews.apache.org/r/8128/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Charles Reiss > >
