Monitoring data become unavailable after updating configuration

[tnagao7]

GlassFish Version (and build number)

GlassFish 7.0.25

JDK version

OpenJDK 21.0.3

OS

Red Hat Enterprise Linux 8

Database

none

Problem Description

The monitoring data become unavailable after updating configuration related to the monitoring target.

Steps to reproduce

1. Enable the http-service monitoring and start the cluster.

$ asadmin create-cluster cluster1
$ asadmin create-local-instance --cluster cluster1 cluster1-1
$ asadmin enable-monitoring --modules http-service --target cluster1
$ asadmin start-cluster cluster1

2. Send requests and see the monitoring data. We can see that the monitoring data are updated.

$ curl http://localhost:28080/
$ curl http://localhost:28080/
$ curl http://localhost:28080/

$ curl -k -H "Accept: application/json" https://localhost:24848/monitoring/domain1/cluster1-1/network/http-listener-1/connection-queue | jq
-- snip --
      "counttotalconnections": {
        "unit": "count",
        "lastsampletime": 1750835156227,
        "name": "CountTotalConnections",
        "count": 3,
        "description": "Total number of connections that have been accepted",
        "starttime": 1750834844252
      },

3. Change any network-related configuration.

$ asadmin set cluster1-config.network-config.protocols.protocol.http-listener-1.http.timeout-seconds=30

4. Send requests and see the monitoring data again. We can see that the monitoring data are not updated. (In the following example, "count" is expected to be "3" but actually "0".)

$ curl http://localhost:28080/
$ curl http://localhost:28080/
$ curl http://localhost:28080/

$ curl -k -H "Accept: application/json" https://localhost:24848/monitoring/domain1/cluster1-1/network/http-listener-1/connection-queue | jq
-- snip --
      "counttotalconnections": {
        "unit": "count",
        "lastsampletime": 1750835275847,
        "name": "CountTotalConnections",
        "count": 0,
        "description": "Total number of connections that have been accepted",
        "starttime": 1750835275840
      },

Impact of Issue

After updating configuration, we cannot continue monitoring without restarting the cluster.