Monitor Your Deployment

Monitor mongot Pod and Resource Status

The most direct signal of mongot availability is the state of its Pod and the status of the MongoDBSearch resource.

Check the pod state and restart count:

kubectl get pods -n <namespace> --context <context> | grep search

A healthy mongot pod reports a Running state with all containers ready and a restart count that doesn't increase. A rising restart count can indicate memory pressure or a configuration error. To learn more, see Troubleshoot mongot Deployment.

Inspect the MongoDBSearch resource status and conditions:

kubectl describe mongodbsearch <MongoDBSearch.metadata.name> \
  -n <namespace> --context <context>

Enable Prometheus Metrics

By default, the mongot Prometheus metrics endpoint is disabled. Add the spec.observability.prometheus block to your MongoDBSearch resource to enable it. An empty block enables the endpoint on the default port 9946.

spec:
  observability:
    prometheus: {}

To use a non-default port, set spec.observability.prometheus.port:

spec:
  observability:
    prometheus:
      port: 9090

The endpoint is exposed inside the pod and is reachable through the mongot container's pod IP on the configured port. The Kubernetes Operator doesn't authenticate the metrics endpoint.

Forward Metrics to Ops Manager

The metrics forwarder sends mongot metrics to Ops Manager so that you can view them alongside your other deployment metrics. The Kubernetes Operator deploys the forwarder as a separate Kubernetes Deployment that scrapes the mongot metrics endpoint and pushes the metrics to Ops Manager.

The metrics forwarder requires Ops Manager 8.0.25 or later and does not support Cloud Manager.

To configure the forwarder, set the following fields under spec.observability.metricsForwarder on the MongoDBSearch resource:

• mode controls whether the Kubernetes Operator deploys the forwarder. Set it to one of the following values:

  • auto (default) automatically deploys the forwarder for an internal MongoDB source, and for an external source when you set opsManager.

  • enabled always deploys the forwarder.

  • disabled never deploys the forwarder.

• opsManager.projectConfigMapRef names the ConfigMap that holds the Ops Manager project configuration. Populate the ConfigMap with the following keys:

  • baseUrl: the URL of your Ops Manager instance.

  • projectName: the name of the Ops Manager project.

  • orgId: the ID of the Ops Manager organization.

• opsManager.agentCredentials names the Secret that holds the Ops Manager API key. Populate the Secret with the following keys:

  • publicKey: the public key of the Ops Manager API key.

  • privateKey: the private key of the Ops Manager API key.

• resourceRequirements sets the CPU and memory for the forwarder container.

If you don't set opsManager, the Kubernetes Operator derives the Ops Manager connection details from the source MongoDB resource. You must set opsManager for an external MongoDB source.

spec:
  observability:
    metricsForwarder:
      mode: enabled
      opsManager:
        projectConfigMapRef:
          name: my-om-project-config
        agentCredentials:
          name: my-om-agent-api-key

The Kubernetes Operator reports the forwarder status under status.metricsForwarder on the MongoDBSearch resource.

Configure Log Verbosity

mongot writes structured JSON logs to its container's stdout and stderr streams, so you can read them with standard Kubernetes log tools:

kubectl logs statefulset/<MongoDBSearch.metadata.name>-search-0

To stream logs in real time, add the --follow flag. To view logs from a previous container instance after a restart, add the --previous flag.

To change the log verbosity, set spec.logLevel on the MongoDBSearch resource. Valid values are TRACE, DEBUG, INFO, WARN, and ERROR. If omitted, mongot logs at INFO.

spec:
  logLevel: DEBUG

mongot doesn't rotate its own logs. In a Kubernetes deployment, the container runtime (containerd or CRI-O on most distributions) handles log rotation at the node level. If you forward logs to an external aggregator, configure retention in that system rather than at the mongot process.

Health Check Endpoint

mongot exposes an HTTP health server on port 8080 inside the pod with two endpoints:

The endpoints reflect mongot startup, liveness, and the initialization of index replication. The endpoints cannot be disabled.

The Kubernetes Operator wires /health and /ready to the mongot container's liveness and readiness probes automatically. You typically don't need to override these probes; kubectl describe pod <mongot-pod> shows the probe configuration applied by the operator.

FTDC Diagnostics

mongot writes Full-Time Diagnostic Data Capture (FTDC) files to the persistent volume mounted at the pod's data path, under a diagnostic.data/ subdirectory. FTDC is enabled by default and is intended for MongoDB Support to triage performance and stability issues.

FTDC files include higher-cardinality data than the Prometheus surface, including per-index sizing, latency breakdowns for indexing and querying, request catalog statistics, serialization latency, and Lucene performance counters.

To collect FTDC files for a support case, copy the diagnostic.data/ directory from the mongot pod with kubectl cp:

kubectl cp <mongot-pod>:/mongot/data/diagnostic.data ./mongot-ftdc

The in-pod mount path is fixed at /mongot/data regardless of the spec.clusters[].persistence configuration. The persistence settings control the PVC size and storage class, not the mount location.

Managed Envoy Observability

When you set spec.clusters[].loadBalancer.managed, the Kubernetes Operator provisions an Envoy L7 proxy that fronts the mongot pods. The Envoy container exposes two observability surfaces:

• Admin endpoint on port 9901 for statistics and runtime control (log level, listener draining). The operator restricts the admin interface to an allowlist of paths: /ready, /stats*, /drain_listeners, and /logging*. Use this endpoint for ad-hoc troubleshooting with kubectl port-forward; it is not intended for external exposure.

• Prometheus metrics on the Envoy admin port at the standard /stats/prometheus path. Scrape this endpoint alongside the mongot metrics to get end-to-end visibility into request flow, upstream health, and connection state for the search traffic plane.

The operator reports the high-level Envoy status under status.loadBalancer.phase on the MongoDBSearch resource:

kubectl get mongodbsearch <name> -o jsonpath='{.status.loadBalancer.phase}'

The phase reflects the operator-managed Envoy Deployment lifecycle (for example, Pending, Running, Failed).

Learn More

• Troubleshoot mongot Deployment
• mongot Deployment Limitations
• spec.observability.prometheus
• spec.logLevel