> On April 21, 2014, 9:47 p.m., Vinod Kone wrote: > > src/slave/slave.cpp, line 738 > > <https://reviews.apache.org/r/20502/diff/1/?file=562543#file562543line738> > > > > Why 16 seconds? Seems too low of a value? How about 10 min like we did > > for status updates? > > > > Either way pull this out to a constant (e.g., SLAVE_RETRY_INTERVAL_MAX). > > Jiang Yan Xu wrote: > We do want them to register prior to the 75 second time out right?
oh yea thats right. s/10min/75s/. Ideally we should extract out the slave ping timeout and max ping retries into a common/constants.hpp header so that it can be used by both the master and slave. maybe a TODO for now. > On April 21, 2014, 9:47 p.m., Vinod Kone wrote: > > src/slave/slave.cpp, lines 729-740 > > <https://reviews.apache.org/r/20502/diff/1/?file=562543#file562543line729> > > > > Instead of doing this way, can we just add the jitter the first time > > doReliableRegistration() is called (e.g., in in detected()) > > > > That way the code here could be simplified (similar to what we did in > > status update manager) and easy to reason about? > > Jiang Yan Xu wrote: > That way how do I express the randomness between retries? (which I think > would be helpful in this case because we don't want everyone to back off too > much and thus delay the recovery time after a failover.) I don't think I follow. Why would everyone backoff too much? - Vinod ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/20502/#review40953 ----------------------------------------------------------- On April 20, 2014, 9:49 p.m., Jiang Yan Xu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/20502/ > ----------------------------------------------------------- > > (Updated April 20, 2014, 9:49 p.m.) > > > Review request for mesos, Ben Mahler and Vinod Kone. > > > Bugs: MESOS-982 > https://issues.apache.org/jira/browse/MESOS-982 > > > Repository: mesos-git > > > Description > ------- > > See summary. > > > Diffs > ----- > > src/slave/slave.hpp 1e9879582315fabb76659e9f8eb03f90188fbfa0 > src/slave/slave.cpp d6ec87c2232c2172f471ae30711b5da1c7050570 > > Diff: https://reviews.apache.org/r/20502/diff/ > > > Testing > ------- > > make check. Also tested the random backoff logic separately. > > > Thanks, > > Jiang Yan Xu > >
