> On Nov. 22, 2016, 1:19 p.m., Benjamin Mahler wrote: > > 3rdparty/libprocess/src/process.cpp, lines 2476-2477 > > <https://reviews.apache.org/r/53997/diff/1/?file=1568675#file1568675line2476> > > > > Do you want an else if here to match the close logic? Otherwise, maybe > > add back the newline here?
Matching the `close` logic sounds like a good idea. To completely match the logic (and hence the assumption that the set of temporary vs persistent sockets do not overlap), I'd also want to flip the if-statements around: swap persistent ones first, else temporary ones. - Joseph ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/53997/#review156632 ----------------------------------------------------------- On Nov. 22, 2016, 2:01 p.m., Joseph Wu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/53997/ > ----------------------------------------------------------- > > (Updated Nov. 22, 2016, 2:01 p.m.) > > > Review request for mesos, Benjamin Mahler and Joris Van Remoortere. > > > Bugs: MESOS-6621 > https://issues.apache.org/jira/browse/MESOS-6621 > > > Repository: mesos > > > Description > ------- > > This fixes some potential CHECK failures when a libprocess process > has (1) SSL downgrade enabled and (2) temporary and persistent > connections open with the same remote address. The second point is > only possible if messages are to a remote address without a persistent > connection and then a persistent connection is created. > > The SSL downgrade path was only checking if the address of a socket > matched when performing the downgrade. The code must also check to > see if the socket itself matches. > > > Diffs > ----- > > 3rdparty/libprocess/src/process.cpp > 84971fa5151991c51e78abdbc736c719e30588f1 > > Diff: https://reviews.apache.org/r/53997/diff/ > > > Testing > ------- > > See related ticket for the clunky unit test. Ran that test in repetition. > > make check > > > Thanks, > > Joseph Wu > >
