----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/28807/#review64744 -----------------------------------------------------------
Ship it! Ship It! - Andrew Onischuk On Dec. 8, 2014, 8:12 p.m., Dmitro Lisnichenko wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/28807/ > ----------------------------------------------------------- > > (Updated Dec. 8, 2014, 8:12 p.m.) > > > Review request for Ambari and Andrew Onischuk. > > > Bugs: AMBARI-8591 > https://issues.apache.org/jira/browse/AMBARI-8591 > > > Repository: ambari > > > Description > ------- > > Ambari deploy's Yarn RM HA and does not enable work preserving restarts. > Which control how many time am can restart are not set accounting for HA > cluster. > mapred-site.xml -> mapreduce.am.max-attempts (defaults to 2) > tez-site.xml -> tez.am.max.app.attempts (defaults to 2) > yarn-site.xml -> yarn.resourcemanager.am.max-attempts (defaults to 2) > In gsInstaller we set this all to 20 which might be over kill but i think > when yarn HA is enabled we should set these to a better value. Yarn team > should provide input on what we should set them if HA is on and if they > should be different if WRP is on as well. > > > Diffs > ----- > > > ambari-server/src/main/resources/stacks/HDP/2.2/services/YARN/configuration/yarn-site.xml > c480566 > > Diff: https://reviews.apache.org/r/28807/diff/ > > > Testing > ------- > > manual verification > > > Thanks, > > Dmitro Lisnichenko > >
