It seem that the system at some point failed and it never able to recover. the only resolution at the moment is to restart datadog-agent. I am suspecting the bug is caused by #432 where during the clean up this.connection set to null but the instance never re-initialize the connection.
I am not familiar with Java as i am not java developer, but i am thinking will replacing connection.getMBeanInfo to this.getConnection().getMBeanInfo fix this issue. but the getConnection() probably need to set this.connection too
2024-04-16 05:15:51 NZST | JMX | INFO | Instance | Refreshing bean list for jboss_wildfly
2024-04-16 05:16:51 NZST | JMX | WARN | App | Unable to collect metrics or refresh bean list.: java.util.concurrent.CancellationException
2024-04-16 05:16:51 NZST | JMX | WARN | App | Instance jboss_hcn didn't return any metrics. Maybe the server got disconnected ? Trying to reconnect.
2024-04-16 05:16:51 NZST | JMX | WARN | Instance | Cannot get attributes or class name for bean jboss.as:subsystem=infinispan,cache-container=hibernate,cache=hcn.ear/hcn-codetabledata-jpa#CodeTableJPAUnit.nz.govt.cyf.hcn.server.entity.codetables.PlanDurationType:
java.lang.NullPointerException: Cannot invoke "org.datadog.jmxfetch.Connection.getMBeanInfo(javax.management.ObjectName)" because "this.connection" is null
at org.datadog.jmxfetch.Instance.getMatchingAttributes(Instance.java:573)
at org.datadog.jmxfetch.Instance.getMetrics(Instance.java:482)
at org.datadog.jmxfetch.MetricCollectionTask.call(MetricCollectionTask.java:26)
at org.datadog.jmxfetch.MetricCollectionTask.call(MetricCollectionTask.java:8)
at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:264)
at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
at java.base/java.lang.Thread.run(Thread.java:840)
Tested on:
========
JMXFetch
Information
runtime_version : 17.0.9
version : 0.48.0
Initialized checks
jboss_wildfly
- instance_name: jboss_hcn
metric_count: 0
service_check_count: 0
message: Unable to instantiate or initialize instance jboss_hcn for an unknown reason.null
status: ERROR
- instance_name: jboss_hcn
metric_count: 0
service_check_count: 0
message: Unable to instantiate or initialize instance jboss_hcn for an unknown reason.null
status: ERROR
Failed checks
one side effect happen is that i see two instances when it start to fail. but if it is running normally, it only show one instance
It seem that the system at some point failed and it never able to recover. the only resolution at the moment is to restart datadog-agent. I am suspecting the bug is caused by #432 where during the clean up this.connection set to null but the instance never re-initialize the connection.
I am not familiar with Java as i am not java developer, but i am thinking will replacing
connection.getMBeanInfotothis.getConnection().getMBeanInfofix this issue. but the getConnection() probably need to set this.connection tooTested on:
one side effect happen is that i see two instances when it start to fail. but if it is running normally, it only show one instance