----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/70116/#review213404 -----------------------------------------------------------
src/master/master.cpp Lines 2261 (patched) <https://reviews.apache.org/r/70116/#comment299342> Why not use `updateOperationMetrics(OPERATION_ERROR, 1);` here? src/master/master.cpp Lines 9223-9224 (patched) <https://reviews.apache.org/r/70116/#comment299343> s/agent/master/ Yea I think we can crash here. src/master/master.cpp Lines 11672-11673 (patched) <https://reviews.apache.org/r/70116/#comment299353> In the case of operation status update retries, this will be unnecessary work (i.e. we will be decrementing and then incrementing the same metric). Perhaps we should enclose this in a conditional which checks for that equality? It will look a bit strange next to the below conditional, but as the comment indicates we need to rethink this deduplication: ``` // TODO(gkleiman): Revisit the de-duplication logic (MESOS-8441) - if two // different terminal statuses arrive, we could end up with different states // in `latest_status` and the front of statuses list. if (operation->statuses().empty() || *(operation->statuses().rbegin()) != status) { operation->add_statuses()->CopyFrom(status); } ``` src/master/metrics.hpp Lines 63-67 (patched) <https://reviews.apache.org/r/70116/#comment299354> Is this comment accurate? Looks like there is indeed an `operations_unreachable` metric below. I think we could probably skip the metrics for such states as the comment suggests. - Greg Mann On March 4, 2019, 5:01 p.m., Benno Evers wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/70116/ > ----------------------------------------------------------- > > (Updated March 4, 2019, 5:01 p.m.) > > > Review request for mesos, Gastón Kleiman, Greg Mann, and Joseph Wu. > > > Bugs: MESOS-8241 > https://issues.apache.org/jira/browse/MESOS-8241 > > > Repository: mesos > > > Description > ------- > > This commit adds the following additional metrics > to the master: > > - master/operations_pending > - master/operations_recovering > - master/operations_finished > - master/operations_failed > - master/operations_error > - master/operations_dropped > - master/operations_unreachable > - master/operations_gone_by_operator > > Unit tests are added in the subsequent commit. > > > Diffs > ----- > > src/master/master.hpp aceab3455adfdf7ff6e168e033316997a4d36fb5 > src/master/master.cpp 665c1c7979a5ae4ecec0d5a68e59c5419049a4d5 > src/master/metrics.hpp 4495e65b6bb11f7236335a702c4f61e7c3f9b0aa > src/master/metrics.cpp 4dd73fb18a06ce8f75c4c1435dba84ade123bee9 > > > Diff: https://reviews.apache.org/r/70116/diff/1/ > > > Testing > ------- > > > Thanks, > > Benno Evers > >
