[ https://issues.apache.org/jira/browse/GEODE-681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15110944#comment-15110944 ]
ASF subversion and git services commented on GEODE-681: ------------------------------------------------------- Commit 6e52f7a3855f045076ea32ae8d913d2ee30ad33e in incubator-geode's branch refs/heads/develop from [~upthewaterspout] [ https://git-wip-us.apache.org/repos/asf?p=incubator-geode.git;h=6e52f7a ] GEODE-681: Measuring event queue size without using the stat for gfsh The eventQueueSize stat has issues with concurrent events during start and stop that are leading it to be inaccurate. To make gfsh and jmx correct, measuring the queue size directly from the underlying region size. Adding unit tests for the size methods. > Stopping and restarting a gateway sender results in incorrect event queue size > ------------------------------------------------------------------------------ > > Key: GEODE-681 > URL: https://issues.apache.org/jira/browse/GEODE-681 > Project: Geode > Issue Type: Bug > Reporter: Dan Smith > Assignee: Dan Smith > > The gateway sender maintains a stat that tracks the number of queued events > in there region. > It does so by reading the number of events in the region after creating the > queue and incrementing the stat by that number, and by having listeners on > the region that also update the stat when entries are added or removed. > Unfortunately, that approach can result in incorrect values for the stat > because events can be double counted by the listener and the code that reads > the size of the region. -- This message was sent by Atlassian JIRA (v6.3.4#6332)