[ https://issues.apache.org/jira/browse/CASSANDRA-6546?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13861500#comment-13861500 ]
Jason Harvey commented on CASSANDRA-6546: ----------------------------------------- Looking through the source I don't see how the metric could be incorrect. It's pretty cut and dry, just a activeSocketSessions.size(); . Perhaps sockets are being left open erroneously :/ ? > connectedThriftClients metric appears to be incorrect after thrift stop/start > ----------------------------------------------------------------------------- > > Key: CASSANDRA-6546 > URL: https://issues.apache.org/jira/browse/CASSANDRA-6546 > Project: Cassandra > Issue Type: Bug > Reporter: Jason Harvey > Priority: Minor > > Disabling and then re-enabling thrift resulted in this stat reporting double > the number of connected sockets on the service. > Alternatively, if the metric is accurate, then disabling and then re-enabling > thrift may be resulting in a bunch of unused FDs? > Reproducing is easy. Have a handful of thrift clients, disablethrift, wait a > few minutes, re-enable thrift, have your client apps reconnect, then compare > actual connected sockets (via netstat or lsof or what have you) to the value > from this MBean ( > org.apache.cassandra.metrics:type=Client,name=connectedThriftClients ). -- This message was sent by Atlassian JIRA (v6.1.5#6160)