[ https://issues.apache.org/jira/browse/CASSANDRA-10175?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Stefan Miklosovic updated CASSANDRA-10175: ------------------------------------------ Resolution: Fixed Status: Resolved (was: Open) This seems to not happen anymore after CASSANDRA-12585 was introduced. I tried to turn off a node and JMX metrics just stopped to be collected. After I started that node, it just continued to report them. There was try-catch added here (1) which catches errors which happen here (2). (1) https://github.com/apache/cassandra/blob/trunk/tools/stress/src/org/apache/cassandra/stress/report/StressMetrics.java#L237-L244 (2) https://github.com/apache/cassandra/blob/trunk/tools/stress/src/org/apache/cassandra/stress/util/JmxCollector.java#L112 > cassandra-stress should be tolerant when a remote node shutdown > ---------------------------------------------------------------- > > Key: CASSANDRA-10175 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10175 > Project: Cassandra > Issue Type: Improvement > Components: Tool/stress > Reporter: Alan Boudreault > Assignee: Stefan Miklosovic > Priority: Normal > Labels: stress > Fix For: 5.x > > > Currently, if we start a stress session with 3 nodes and shutdown one node, > stress will crash. It is caused by the JMX connection lost on the node, which > is use to collect some gc stats IIRC. > backtrace: https://gist.github.com/aboudreault/6cd82bb0acc681992414 > Stress should handle that jmx connection lost in a better way so the session > could continue. Ideally, it should try to *reconnect* to JMX if the node is > back online? -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org