Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
Description
When an agent looses a JMX connection to a remote service it does not try to recover it. Instead just reports 0's for GC values.
The problem is that it checks connection for null and only if true the reconnect happens. In the failing scenario, the connection is established and works for awhile than the connection fails due to a long GC in the service. The connection object is stale (not null) at this point and reconnect does not happen.
When a JMX failure occurs set the connection object to null forcing a subsequent reconnect. When the JMX fails to fetch GC info the GC collection should set GC values to -1. This info will be sent to Duccmon where the failure can be visualized. Perhaps changing color of GC value to red could alert a user.