Java Services Metrics
MTG CLM, ACME, AEC, CMP, EST, SCEP, CARA WS, and the Revocation Information Service expose metrics for CPU, disk space, JVM memory, HTTP requests, garbage collection and threads. MTG CLM, ACME, and CARA WS also expose database connection-pool metrics.
Component labels
Each series includes application and host labels.
host identifies the application instance.
| Component | application label |
|---|---|
CLM |
|
ACME |
|
AEC |
|
CMP |
|
EST |
|
SCEP |
|
CARA WS |
|
Revocation Information Service |
|
CPU, disk, process, and memory
These measurements are gauges. Memory labels identify the area, either heap or nonheap, and the memory-pool id. Disk measurements identify the monitored path.
| Prometheus sample | Unit | Extra labels | Meaning |
|---|---|---|---|
|
Ratio, normally 0-1 |
None |
Recent CPU usage reported by the operating system available to the JVM. |
|
Ratio, normally 0-1 |
None |
Recent CPU usage of the application JVM process. |
|
Unix epoch seconds |
None |
JVM process start time. |
|
Bytes |
|
Usable filesystem space for the monitored path. |
|
Bytes |
|
Total filesystem capacity for that path. |
|
Bytes |
|
Used memory in each JVM memory pool. |
|
Bytes |
|
Memory committed for the JVM to use in each pool. |
|
Bytes |
|
Maximum pool capacity; |
HTTP requests
Completed-request metrics use method, uri, status, outcome, exception, and error labels. The uri label normally contains the route pattern. Active-request metrics use the labels available when the request starts.
| Prometheus sample | Type and unit | Meaning |
|---|---|---|
|
Cumulative count |
Completed requests per label combination. |
|
Cumulative seconds |
Total duration of those completed requests. |
|
Gauge, seconds |
Largest request duration in the timer’s rolling maximum window. |
|
Current count |
Requests currently in progress. |
|
Current seconds |
Sum of elapsed durations of requests currently in progress. |
|
Gauge, seconds |
Longest elapsed duration among requests currently in progress. |
Garbage collection
GC timers use gc, action, and cause labels to identify the collector and collection event. jvm_gc_cpu_time_ns_total applies to CLM, ACME, AEC, CMP, EST, SCEP, and the Revocation Information Service on JVMs that supply GC CPU time.
| Prometheus sample | Type and unit | Extra labels | Meaning or condition |
|---|---|---|---|
|
Gauge, ratio |
None |
Recent GC pause time divided by the monitoring interval, approximating the fraction of time spent in GC. |
|
Gauge, bytes |
None |
Maximum size of long-lived heap pools. |
|
Gauge, bytes |
None |
Long-lived heap memory used after reclamation; updated on relevant GC events. |
|
Counter, bytes |
None |
Positive allocation-pool growth between GC events. This is a GC-derived estimate. |
|
Counter, bytes |
None |
Bytes promoted to old-generation memory during garbage collection. |
|
Cumulative count |
|
Recorded GC pauses. |
|
Cumulative seconds |
|
Total recorded GC pause duration. |
|
Gauge, seconds |
|
Rolling maximum GC pause duration. |
|
Cumulative count |
|
Recorded concurrent GC phases. |
|
Cumulative seconds |
|
Total recorded concurrent-phase duration. |
|
Gauge, seconds |
|
Rolling maximum concurrent-phase duration. |
|
Counter, nanoseconds |
None |
Accumulated CPU time used for garbage collection. |
JVM threads
The state label identifies new, runnable, blocked, waiting, timed-waiting, or terminated threads.
| Prometheus sample | Type and unit | Extra labels | Meaning |
|---|---|---|---|
|
Gauge, threads |
None |
Current live threads, including daemon and non-daemon threads. |
|
Gauge, threads |
None |
Current live daemon threads. |
|
Gauge, threads |
None |
Peak live thread count since JVM start or the last peak reset. |
|
Counter, threads |
None |
Total application threads started by the JVM. |
|
Gauge, threads |
|
Threads in each state: |
Database connection pools
CLM and ACME
These metrics apply to CLM and ACME. Each series has a pool label.
| Prometheus sample | Type and unit | Meaning |
|---|---|---|
|
Gauge, connections |
Total connections in the pool. |
|
Gauge, connections |
Connections currently borrowed. |
|
Gauge, connections |
Idle connections. |
|
Gauge, threads |
Threads waiting for a connection. |
|
Gauge, connections |
Configured maximum pool size. |
|
Gauge, connections |
Configured minimum idle connections. |
|
Counter, timeouts |
Connection-acquisition timeouts. |
|
Cumulative count |
Recorded connection acquisitions. |
|
Cumulative seconds |
Total acquisition wait time. |
|
Gauge, seconds |
Rolling maximum acquisition wait time. |
|
Cumulative count |
Recorded connection creations. |
|
Cumulative seconds |
Total connection creation time. |
|
Gauge, seconds |
Rolling maximum connection creation time. |
|
Cumulative count |
Recorded completed connection uses. |
|
Cumulative seconds |
Total time connections were borrowed. |
|
Gauge, seconds |
Rolling maximum connection usage time. |
CARA WS
CARA WS exposes Tomcat JDBC connection-pool metrics.
Each series includes a name label identifying the datasource, in addition to application and host.
All four metrics are gauges measured in connections.
| Prometheus sample | Type and unit | Meaning |
|---|---|---|
|
Gauge, connections |
Connections currently borrowed from the pool. |
|
Gauge, connections |
Established connections available for use. |
|
Gauge, connections |
Configured maximum number of connections that can be active at the same time. |
|
Gauge, connections |
Configured minimum number of idle connections. |
HTTP metrics measure requests and response processing time. Use application audit records to examine individual certificate operations.