enhanced cluster: When inter-broker heartbeat resumes, an indoubt broker should be removed from suspect list

glassfishrobot commented 11 years ago

The default setting of cluster service (inter-broker, UDP) heartbeat timeout in a MQ enhanced cluster are imq.cluster.heartbeat.interval=2 (in seconds) imq.cluster.heartbeat.threshold=3

After this timeout, a broker is put on the "suspected" list for the HA monitor service to monitor its heartbeat to the HA database. Network fluctuation or load between brokers can make a broker in the cluster susceptible as "suspected". While monitored by the HA monitor service, if the inter-broker heartbeat resumes, the "suspected" broker should be removed from the "suspected" list. This reduces the chance of false takeover and broker restart (as result of a false takeover detected) if the heartbeat to the HA database also timeout

glassfishrobot commented 6 years ago

Issue Imported From: https://github.com/javaee/openmq/issues/241
Original Issue Raised By:@glassfishrobot
Original Issue Assigned To: @amykang2020

glassfishrobot commented 11 years ago

@glassfishrobot Commented Reported by @amykang2020

glassfishrobot commented 11 years ago

@glassfishrobot Commented @amykang2020 said: The workaround is to increase the inter-broker heartbeat setting imq.cluster.heartbeat.interval (in seconds) imq.cluster.heartbeat.threshold

glassfishrobot commented 7 years ago

@glassfishrobot Commented This issue was imported from java.net JIRA MQ-241

eclipse-ee4j / openmq

enhanced cluster: When inter-broker heartbeat resumes, an indoubt broker should be removed from suspect list #241