----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/72368/#review220377 -----------------------------------------------------------
Fix it, then Ship it! src/slave/slave.cpp Lines 1077 (patched) <https://reviews.apache.org/r/72368/#comment308736> Please make sure that you have dealt with `SlaveTest.DrainInfoInAPIOutputs` that expect an empty agent to be draining. I've seen this at some point but completely forgot about that; thankfully, this test showed up in the CI results. - Andrei Sekretenko On April 17, 2020, 6:04 p.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/72368/ > ----------------------------------------------------------- > > (Updated April 17, 2020, 6:04 p.m.) > > > Review request for mesos, Andrei Sekretenko and Benjamin Mahler. > > > Bugs: MESOS-10118 > https://issues.apache.org/jira/browse/MESOS-10118 > > > Repository: mesos > > > Description > ------- > > Previously, when the agent had no tasks or operations and > received a `DrainSlaveMessage`, it would checkpoint the > `DrainConfig` to disk, implicitly placing it into a "draining" > state indefinitely. This patch updates the agent's handler to > avoid checkpointing anything to disk in this case. > > > Diffs > ----- > > src/slave/slave.cpp 1a32c813eeacf21b903db3f9602d034e7fb085c0 > > > Diff: https://reviews.apache.org/r/72368/diff/1/ > > > Testing > ------- > > Ran the test in https://reviews.apache.org/r/72364/ with this patch applied > and verified that the final task reached TASK_RUNNING. > > > Thanks, > > Greg Mann > >
