> On Oct. 24, 2017, 4:58 p.m., Benjamin Mahler wrote: > > What goes wrong with this issue? Would be good to clarify that in the > > description. > > > > Have you considered a test?
I did consider a test, but as it is right now I think it's rather difficult. If I run into another issue with recovered agents state, I'll consider more strongly. - Michael ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/63232/#review189119 ----------------------------------------------------------- On Nov. 5, 2017, 7:24 p.m., Michael Park wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/63232/ > ----------------------------------------------------------- > > (Updated Nov. 5, 2017, 7:24 p.m.) > > > Review request for mesos, Benjamin Mahler and Greg Mann. > > > Bugs: MESOS-7851 > https://issues.apache.org/jira/browse/MESOS-7851 > > > Repository: mesos > > > Description > ------- > > The intention is for us to work with resources in the `post-reservation- > refinement` format in the master memory, and store them in the master > registry or agent checkpoint in the `pre` format in order to support > downgrades. In this case, we correctly store the `pre` format resources > in the master registry, but we don't convert them back when we read it > out of the registry. This patch converts the resources back to `post` > when we read from the registry. > > This simply addresses a tech-debt. The `authorizeResource` function > for example currently handles both formats because it has to account > for the old format that is stored in the recovered agents. > > > Diffs > ----- > > src/master/master.cpp 49bc50ead75adba82a7d8de23a87b06ccd269c48 > > > Diff: https://reviews.apache.org/r/63232/diff/2/ > > > Testing > ------- > > `make check` > > > Thanks, > > Michael Park > >
