IBM MQ OpenTelemetry metrics and attributes reference
This reference covers the IBM MQ metrics and attributes collected via OpenTelemetry: the metrics the mq-metric-samples exporter publishes and the collector forwards to New Relic. It applies to both the Linux host and Kubernetes deployments; the metric shape is identical.
Metric collection
All metrics keep the raw Prometheus shape the exporter emits: underscore names (ibmmq_*) and the raw qmgr, queue, and channel labels, plus the target.name identity attribute and optional cluster.name tag the collector attaches. New Relic's IBM MQ entity synthesis depends on this shape, so the collector forwards the names and labels unchanged.
Important
Enable MQI statistics on each queue manager, or the throughput and timing metrics stay empty. Queue-manager health, queue depth, and channel status report out of the box, but the MQI activity metrics (the queue throughput and queue manager MQI activity tables — ibmmq_queue_mqput_count, ibmmq_queue_average_queue_time_seconds, and so on) require statistics collection to be on, which is off by default. Enable it once per queue manager with MQSC — for example ALTER QMGR STATMQI(ON) STATQ(ON), then ALTER QLOCAL(...) STATQ(ON) for any queue you want per-queue stats on (or rely on the queue-manager STATQ default). Without this, the exporter is healthy and the entity appears, but those panels render blank. See the mq-metric-samples documentation for the exact statistics each metric depends on.
Tip
Status metrics report numeric enum values, not 1/0 — see the status-code reference below before alerting on them. Counters are converted to delta temporality by the collector, so use sum() / rate() over a window rather than latest() when charting them.
The tables below group the OpenTelemetry metrics by area. To list every metric actually reporting in your account, run:
FROM Metric SELECT uniques(metricName,2000)WHERE metricName LIKE'ibmmq_%' SINCE 1day ago
Metrics
Metric
Description
ibmmq_qmgr_status
Queue manager run state (2 = running).
ibmmq_qmgr_command_server_status
Command server run state (2 = running).
ibmmq_qmgr_channel_initiator_status
Channel initiator (the queue-manager process that starts and manages MQ channels) run state (2 = running).
ibmmq_qmgr_connection_count
Current number of connections to the queue manager.
ibmmq_qmgr_uptime
Queue manager uptime, in seconds.
ibmmq_qmgr_active_listeners
Number of active listeners.
ibmmq_qmgr_active_services
Number of active services.
ibmmq_qmgr_concurrent_connections_high_water_mark
Peak concurrent connections.
Metric
Description
ibmmq_qmgr_cpu_load_one_minute_average_percentage
1-minute CPU load average for the queue manager (also _five_ and _fifteen_minute_).
ibmmq_qmgr_user_cpu_time_percentage
User-space CPU time consumed by the queue manager.
ibmmq_qmgr_system_cpu_time_percentage
System (kernel) CPU time consumed by the queue manager.
Number of IBM MQ diagnostic dump files (FFST/FDC — written when the queue manager logs an internal error). A rising count signals problems worth investigating.
Metric
Description
ibmmq_queue_depth
Current number of messages on the queue.
ibmmq_queue_attribute_max_depth
Configured maximum queue depth.
ibmmq_queue_oldest_message_age
Age of the oldest message on the queue, in seconds.
ibmmq_queue_uncommitted_messages
Number of uncommitted messages.
ibmmq_queue_time_since_put
Seconds since the last MQPUT to the queue.
ibmmq_queue_time_since_get
Seconds since the last MQGET from the queue.
ibmmq_queue_input_handles
Open input handles (consumers) on the queue.
ibmmq_queue_output_handles
Open output handles (producers) on the queue.
ibmmq_queue_browse_handles
Open browse handles on the queue.
ibmmq_queue_publish_handles
Open publish handles on the queue.
ibmmq_queue_qfile_current_size
Current queue file size on disk.
ibmmq_queue_qfile_max_size
Maximum queue file size on disk.
These counters track IBM MQ's core message API calls per queue: MQPUT / MQPUT1 (put a message onto a queue) and MQGET (get a message from a queue). They require MQI statistics to be enabled on the queue manager (see Metric collection).
These queue-manager-wide counters summarize API activity and failures across all objects — useful for spotting application errors and transaction churn. They require MQI statistics to be enabled (see Metric collection).
Metric
Description
ibmmq_qmgr_interval_mqput_mqput1_total_count
Total MQPUT + MQPUT1 calls in the interval.
ibmmq_qmgr_interval_destructive_get_total_count
Total destructive MQGET calls in the interval.
ibmmq_qmgr_published_to_subscribers_message_count
Messages published to subscribers (publish/subscribe throughput).
IBM MQ status metrics report numeric enum values, not1/0 — and the healthy ("running") value differs by metric (2 for queue-manager-level status, 3 for channel status). Alert on inequality to the running value (for example, ibmmq_qmgr_status != 2):
Metric
Healthy value
Meaning
ibmmq_qmgr_status
2
Queue manager is running
ibmmq_qmgr_command_server_status
2
Command server is running
ibmmq_qmgr_channel_initiator_status
2
Channel initiator is running
ibmmq_channel_status
3
Channel is running
Any other value (or a missing data point) means the component is not in its running state.
Resource attributes
The collector attaches identity and context attributes to every IBM MQ metric. Some are common to all deployments; others depend on whether the collector runs on a Linux host or in Kubernetes.
Attribute
Description
Example
target.name
Stable identity attribute — the first segment of every IBM MQ entity GUID. Injected as the targetName scrape label, then renamed to target.name by the collector.
prod-mq-01, prod-mq-cluster
qmgr
Queue manager name (raw exporter label; entity synthesis keys on it).
QM1
queue
Queue name (on queue-scoped metrics).
APP.ORDERS.IN
channel
Channel name (on channel-scoped metrics).
APP.SVRCONN
cluster.name
Optional grouping tag for queue managers in the same MQ cluster. Not part of the entity GUID; safe to leave unset or change later.
prod-mq
instrumentation.provider
Always opentelemetry for this integration; use it to scope queries to the OpenTelemetry path.
opentelemetry
Attribute
Description
Example
host.id
Identifier of the host running the collector, from the resourcedetection processor (system / ec2 detectors). Links the queue manager to its underlying host entity.
i-1234567890abcdef0
Attribute
Description
Example
k8s_namespace
Namespace of the discovered exporter pod, stamped by the collector's relabel rules.
ibmmq
k8s_pod
Name of the discovered exporter pod.
qm1-0
host.id
Identifier of the node running the pod, from resourcedetection.
ip-10-0-1-23
Metric collection details
Collection frequency
Default interval: 60 seconds (the collector's scrape_interval). It must be greater than or equal to the mq-metric-samples exporter's own pollInterval.
Configurable: adjust IBMMQ_SCRAPE_INTERVAL (Linux) or scrape_interval in the Helm values.yaml (Kubernetes).
Data source
All metrics originate from the mq-metric-samples exporter, which polls each queue manager over PCF and republishes the statistics as Prometheus metrics on an HTTP /metrics endpoint. The collector's prometheus receiver scrapes that endpoint.
Metric types
Gauges — point-in-time values such as ibmmq_qmgr_status, ibmmq_queue_depth, and ibmmq_channel_status. Chart with latest().
Counters — cumulative totals such as ibmmq_queue_mqput_count. The collector's cumulativetodelta processor converts them to delta temporality, so chart with sum() or rate() rather than latest().