Skip to content

Info

License: Free — part of the universal collection tier.

WildFly / JBoss

The wildfly probe monitors WildFly (and JBoss EAP) via the HTTP Management API, collecting JVM heap and GC metrics, Undertow web-container request counters, JTA transaction statistics and per-datasource JDBC connection pool metrics.

Quick start

# probes.d/10-wildfly.yaml — each file under probes.d/ is a YAML array of probes
- name: wildfly
  type: wildfly
  params:
    endpoint: http://localhost:9990
    username: admin
    password: ${secret:wildfly.password}   # OS secret store; inline plaintext is auto-sealed on install

Parameters

Parameter Must set Default Description
endpoint In practice http://localhost:9990 Base URL of the management interface. Example: http://wildfly.example.com:9990
username In practice admin Management user
password In practice - Management user's password. A secret: reference it with ${secret:…}, ${env:…} or ${file:…} rather than writing it in the file
timeout No 10 Request timeout in seconds
interval No 60 Seconds between collections
instance_name No - Stable identity of this server instead of the one derived from the endpoint

Metrics

JVM heap (used, committed, max), Undertow request, error and byte counters, JTA committed and rolled-back transactions, and the active and available connections of each datasource pool. The full list is in the metric reference below.

Operational notes

  • Create a dedicated management user with the Monitor role: bin/add-user.sh -u monitor -p password -g Monitor.
  • The management interface authenticates with HTTP Digest by default (ManagementRealm); the probe answers the Digest challenge, and still works with an interface configured for Basic.
  • For WildFly domain mode, point the endpoint at the domain controller (port 9990).
  • The probe uses the WildFly HTTP Management API (JSON over HTTP), not Jolokia — Jolokia is not required.

Metric reference

Every metric this probe can emit. Metric is the OpenTelemetry name the OTLP, Prometheus and Zabbix outputs derive theirs from. Name is what a Nagios check and the API metrics= filter match. PRTG channel is the label PRTG shows, placeholders filled from the series' tags.

Metric Name PRTG channel Unit Description
senhub.wildfly.up senhub.wildfly.up WildFly Up # 1 when the WildFly Management API responded successfully; 0 otherwise
jvm.memory.heap.used jvm.memory.heap.used WildFly JVM Heap Used B JVM heap memory currently used
jvm.memory.heap.committed jvm.memory.heap.committed WildFly JVM Heap Committed B JVM heap memory committed to the JVM process
jvm.memory.heap.max jvm.memory.heap.max WildFly JVM Heap Max B Maximum JVM heap memory available
wildfly.request.count wildfly.request.count WildFly Requests # Total number of requests processed by Undertow
wildfly.error.count wildfly.error.count WildFly Errors # Total number of error responses from Undertow
wildfly.bytes.sent wildfly.bytes.sent WildFly Bytes Sent B Total bytes sent by Undertow
wildfly.bytes.received wildfly.bytes.received WildFly Bytes Received B Total bytes received by Undertow
wildfly.transaction.committed wildfly.transaction.committed WildFly Transactions Committed # Total number of committed JTA transactions
wildfly.transaction.rolledback wildfly.transaction.rolledback WildFly Transactions Rolled Back # Total number of aborted (rolled back) JTA transactions
wildfly.datasource.connections.active wildfly.datasource.connections.active WildFly DS {datasource} Active Connections # Number of active (in-use) connections in the JDBC datasource pool
wildfly.datasource.connections.available wildfly.datasource.connections.available WildFly DS {datasource} Available Connections # Number of available (idle) connections in the JDBC datasource pool