----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/70613/#review215169 -----------------------------------------------------------
Fix it, then Ship it! Can you file a ticket for this? src/master/allocator/mesos/hierarchical.cpp Lines 2008-2009 (patched) <https://reviews.apache.org/r/70613/#comment301706> "eliminate effect" seems vague, maybe something like "spread out" the effect of the first stage, which tends to allocate from the front of the agent list more so than the back? - Benjamin Mahler On May 9, 2019, 12:14 a.m., Meng Zhu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/70613/ > ----------------------------------------------------------- > > (Updated May 9, 2019, 12:14 a.m.) > > > Review request for mesos, Andrei Sekretenko and Benjamin Mahler. > > > Repository: mesos > > > Description > ------- > > Before this patch, agents are randomized before the 1st > allocation stage (the quota allocation stage) but not in > the 2nd stage. One perceived issue is that resources on > the agents in the front of the queue are likely to be mostly > allocated in the 1st stage, leaving only slices of resources > available for the second stage. Thus we may see consistently > low quality offers for role/frameworks that get allocated first > in the 2nd stage. > > This patch randomizes the agents again before the 2nd stage to > eliminate any effect left by the 1st stage allocation. > > > Diffs > ----- > > src/master/allocator/mesos/hierarchical.cpp > 64a076ddd29711437d539a06bb0470755828cc87 > > > Diff: https://reviews.apache.org/r/70613/diff/1/ > > > Testing > ------- > > make check > > > Thanks, > > Meng Zhu > >
