> On Dec. 15, 2014, 5:14 p.m., Nate Cole wrote: > > ambari-server/src/main/java/org/apache/ambari/server/events/listeners/alerts/AlertAggregateListener.java, > > lines 110-115 > > <https://reviews.apache.org/r/29062/diff/1/?file=792310#file792310line110> > > > > These conditionals (>=) are not consistent with other boundary checks > > (as in metric_alert.py) Looks like the changed port_alert.py also uses >=, > > so we should be consistent (the old nagios didn't use >=, just >) > > Jonathan Hurley wrote: > This came out of a conversation with Jeff about the behavior of the > alerts. Given 10 datanodes; if 5 go down, that would mean that 50% are down. > If the alerts says "WARNING : 0.5" then administrators would expect an alert > to fire at this point in time. However, if the check says .5 > .5 it would > not trigger the alert. > > We're readying this as "% of alerts affected". So if the value is 50, > you're saying if 50% of the alerts are affected, then trigger the state > change. I can change it back, but then it won't work as desired. Thoughts? > > Jonathan Hurley wrote: > Just confirmed this; 5 datanodes down out of 10 should trigger a 50% > state change. >= is what we want to use here.
Then metric_alert.py should also change to be consistent. - Nate ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/29062/#review65132 ----------------------------------------------------------- On Dec. 15, 2014, 4:46 p.m., Jonathan Hurley wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/29062/ > ----------------------------------------------------------- > > (Updated Dec. 15, 2014, 4:46 p.m.) > > > Review request for Ambari and Nate Cole. > > > Bugs: AMBARI-8723 > https://issues.apache.org/jira/browse/AMBARI-8723 > > > Repository: ambari > > > Description > ------- > > Configured cluster with two DataNodes. Modified the percent DNs available > alert to be 10% for warning, 99% for crit. Shutdown one DN and crit alert is > fired. Should be warning since only 50% of DNs are affected (not 99%). > > > Diffs > ----- > > > ambari-server/src/main/java/org/apache/ambari/server/events/AlertDefinitionChangedEvent.java > PRE-CREATION > > ambari-server/src/main/java/org/apache/ambari/server/events/AlertDefinitionDisabledEvent.java > 2fe3ba9 > > ambari-server/src/main/java/org/apache/ambari/server/events/AmbariEvent.java > 04217f1 > > ambari-server/src/main/java/org/apache/ambari/server/events/listeners/alerts/AlertAggregateListener.java > 6d8f34a > > ambari-server/src/main/java/org/apache/ambari/server/events/listeners/alerts/AlertLifecycleListener.java > df95d3d > > ambari-server/src/main/java/org/apache/ambari/server/orm/dao/AlertDefinitionDAO.java > 23de17e > ambari-server/src/main/java/org/apache/ambari/server/orm/dao/AlertsDAO.java > 47d26cb > > ambari-server/src/main/java/org/apache/ambari/server/state/alert/Source.java > 09ddeb3 > > ambari-server/src/main/java/org/apache/ambari/server/state/services/AlertNoticeDispatchService.java > 69f3393 > > ambari-server/src/test/java/org/apache/ambari/server/orm/dao/AlertsDAOTest.java > 31e93e4 > > ambari-server/src/test/java/org/apache/ambari/server/state/alerts/AlertEventPublisherTest.java > 271c078 > > Diff: https://reviews.apache.org/r/29062/diff/ > > > Testing > ------- > > Manual tests between 0, 33%, 50% and 100% for WARNING/CRITICAl aggregate > definitions. New tests added. > > > Thanks, > > Jonathan Hurley > >
