[GEODE-6732] GMSHealthMonitor reports member is not available when self-health check fails

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.10.0
• Component/s: core, membership
• Labels: SmallFeature
• Assignee: Bruce J Schuchardt
• Reporter: Bruce J Schuchardt
• Created: 02/May/19 16:00
• Updated: 26/Sep/19 18:04
• Resolved: 29/May/19 15:43

Description

GMSHealthMonitor does a self-check if another node isn't responding.  If that check fails it still seems to report that the other node is unavailable.

 

} else {
  // if this node can survive an availability check then initiate suspicion about
  // the node that failed the availability check
  if (doTCPCheckMember(localAddress, this.socketPort, false)) {
    membersInFinalCheck.remove(mbr);
    // tell peers about this member and then perform another availability check
    memberSuspected(localAddress, mbr, reason);
    initiateSuspicion(mbr, reason);
    SuspectMembersMessage suspectMembersMessage =
        new SuspectMembersMessage(Collections.singletonList(localAddress),
            Collections
                .singletonList(new SuspectRequest(mbr, "failed availability check")));
    suspectMembersMessage.setSender(localAddress);
    logger.debug("Performing local processing on suspect request");
    processSuspectMembersRequest(suspectMembersMessage);
  }
}
failed = true;

 

Issue Links

• links to: GitHub Pull Request #3621