> On Sept. 12, 2016, 10:52 p.m., Vinod Kone wrote: > > src/master/flags.cpp, line 571 > > <https://reviews.apache.org/r/51021/diff/5/?file=1496749#file1496749line571> > > > > do we want to enforce that `registry_max_agent_age` is >= > > `registry_gc_interval`? otherwise, users might be confused why old agents > > are not being GC'ed?
I considered this. I ended up not doing it because it seemed like additional complexity to explain to users why certain configuration values are illegal. Configuring `registry_max_agent_age` to be smaller than `registry_gc_interval` certainly seems like an usual configuration, but it doesn't seem *wrong* or harmful, so I thought I'd allow it. Let me know if disagree. - Neil ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/51021/#review148615 ----------------------------------------------------------- On Sept. 12, 2016, 1:48 p.m., Neil Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/51021/ > ----------------------------------------------------------- > > (Updated Sept. 12, 2016, 1:48 p.m.) > > > Review request for mesos and Vinod Kone. > > > Bugs: MESOS-5965 > https://issues.apache.org/jira/browse/MESOS-5965 > > > Repository: mesos > > > Description > ------- > > Information about agents that have been marked unreachable is stored > in the registry. Since the number of unreachable agents can grow without > bound, this commit implements a garbage collection scheme. The current > leading master will periodically examine the registry and discard > information about unreachable agents according to two criteria: > `registry_max_agent_age` and `registry_max_agent_count`. The frequency > with which the master examines the registry is controlled by a third > parameter, `registry_gc_interval`. > > > Diffs > ----- > > src/master/constants.hpp cd80dace80968a1f67a8de5b2c112fb1396e26aa > src/master/flags.hpp c6e85303f60387f42b5e187eaedb6a01000f948f > src/master/flags.cpp 19ae6c1c30a1054b64a9585f325bd0bf943af218 > src/master/master.hpp 4992ab0a0bb5babbf6a4fa3e6eff3577590fc879 > src/master/master.cpp 1dcce6cd66804990af238176c61aca03bb5c9471 > src/master/registry.proto 03c896c39458b54b3966ce32c45eaa33c890ed28 > src/tests/partition_tests.cpp f3142ad8d50daafcdb70ad9dbb2772f8ba30db00 > src/tests/registrar_tests.cpp b04fc92c66449f1094212a534899e16d5f6f01bd > > Diff: https://reviews.apache.org/r/51021/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Neil Conway > >
