Monitoring

Health checks

Health

An endpoint is provided to perform health checks on the Nuts node. It reports in a format compatible with Spring Boot’s Health Actuator. The endpoint is available over HTTP:

GET /health

Each component in the health check can have one of the statuses UP, UNKNOWN, or DOWN. The overall status is determined by the lowest common denominator, so if one components is DOWN, the overall system status is DOWN. The overall system statuses UP and UNKNOWN map to HTTP status code 200, and status DOWN maps to status code 503.

Example response when all checks succeeded (formatted for readability):

{
  "status": "UP",
  "details": {
    "crypto.filesystem": {
      "status": "UP"
    },
    "network.auth_config": {
      "status": "UP",
      "details": "no node DID"
    },
    "network.tls": {
      "status": "UP"
    }
  }
}

Example response when one or more checks failed:

{
  "status": "DOWN",
  "details": {
    "network.tls": {
      "status": "DOWN",
      "details": "x509: certificate signed by unknown authority"
    }
  }
}

Note

The provided Docker containers are configured to perform this healthcheck out of the box. However, if the default port (:1323) has been changed or if the /health endpoint has been bound to a different port, the default healthcheck will fail and Docker will mark the container as unhealthy. Override the default healthcheck to solve this.

Status

There’s also a status endpoint to check whether the service has been started. It does not provide any information on the individual engines running as part of the executable. The main goal of the service is to give a YES/NO answer for if the service is running:

GET /status

Returns an “OK” response body with status code 200.

Basic diagnostics

GET /status/diagnostics

Returns the status of the various services in yaml format:

network:
    connections:
        connected_peers:
            - id: d38c6df5-63d2-4b2c-87f4-2e8bbfa5612f
              address: nuts.nl:5555
              nodedid: did:nuts:abc123
        connected_peers_count: 1
    state:
        dag_xor: 6aada4464e380db16d0316e597956fcdaeada0e8f6023be82eeb9c798e1815c6
        stored_database_size_bytes: 106496005
        transaction_count: 9001
vcr:
    credential_count: 7
    issuer:
        issued_credentials_count: 0
        revoked_credentials_count: 0
    verifier:
        revocations_count: 18
vdr:
    did_documents_count: 5
    conflicted_did_documents:
        total_count: 2
        owned_count: 0
status:
    git_commit: d36837bae48b780bfb76134e85b506472fc207a6
    os_arch: linux/amd64
    software_version: master
    uptime: 4h14m12s

If you supply application/json for the Accept HTTP header it will return the diagnostics in JSON format.

Explanation of ambiguous/complex entries in the diagnostics:

vcr.credential_count holds the total number of credentials known to the node (public VCs, and private VCs issued to a DID on the local node)
vcr.issuer.issued_credentials_count holds the total number of credentials issued by the local node
vcr.issuer.revoked_credentials_count holds the total number of revoked credentials issued by the local node
vcr.verifier.revocations_count holds the total number of revoked credentials (public and private VCs)
vdr.conflicted_did_documents.total_count holds the total number of DID documents that are conflicted (have parallel updates). This may indicate a stolen private key
vdr.conflicted_did_documents.owned_count holds the number of conflicted DID documents you control as a node owner

Conflicted DID documents

Conflicted DID documents are the result of parallel updates of a DID document. It can be caused by a race condition during DID API calls, having a cluster of nodes when this is not supported or if a private key has been stolen. To rule out this last cause, every node must make sure their conflicted document count equals 0. There’s an API to get the actual conflicted DID documents. A DID document can be fixed by simply updating it.

Metrics

The Nuts service executable has build-in support for Prometheus. Prometheus is a time-series database which supports a wide variety of services. It also allows for exporting metrics to different visualization solutions like Grafana. See https://prometheus.io/ for more information on how to run Prometheus. The metrics are exposed at /metrics

In order for metrics to be gathered by Prometheus. A job has to be added to the prometheus.yml configuration file. Below is a minimal configuration file that will only gather Nuts metrics:

# my global config
global:
  scrape_interval:     15s # Set the scrape interval to every 15 seconds. Default is every 1 minute.
  evaluation_interval: 15s # Evaluate rules every 15 seconds. The default is every 1 minute.
  # scrape_timeout is set to the global default (10s).

# Load rules once and periodically evaluate them according to the global 'evaluation_interval'.
rule_files:
# - "first_rules.yml"
# - "second_rules.yml"

# A scrape configuration containing exactly one endpoint to scrape:
scrape_configs:
  # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config.
  - job_name: 'nuts'
    metrics_path: '/metrics'
    scrape_interval: 5s
    static_configs:
      - targets: ['127.0.0.1:1323']

It’s important to enter the correct IP/domain and port where the Nuts node can be found!

The Nuts service executable exports the following metric namespaces:

nuts_ contains metrics related to the functioning of the Nuts node
process_ contains OS metrics related to the process
go_ contains Go metrics related to the process
promhttp_ contains metrics related to HTTP calls to the Nuts node’s /metrics endpoint

Network DAG Visualization

All network transactions form a directed acyclic graph (DAG) which helps achieving consistency and data completeness. Since it’s a hard to debug, complex structure, the network API provides a visualization which can be queried from /internal/network/v1/diagnostics/graph. It is returned in the dot format which can then be rendered to an image using dot or graphviz (given you saved the output to input.dot):

dot -T png -o output.png input.dot

Using query parameters start and end it is possible to retrieve a range of transactions. /internal/network/v1/diagnostics/graph?start=10&end=12 will return a graph with all transactions containing Lamport Clock 10 and 11. Both parameters need to be non-negative integers, and start < end. If no value is provided, start=0 and end=inf. Querying a range can be useful if only a certain range is of interest, but may also be required to generate the graph using dot.

CPU profiling

It’s possible to enable CPU profiling by passing the --cpuprofile=/some/location.dmp option. This will write a CPU profile to the given location when the node shuts down. The resulting file can be analyzed with Go tooling:

go tool pprof /some/location.dmp

The tooling includes a help function to get you started. To get started use the web command inside the tooling. It’ll open a SVG in a browser and give an overview of what the node was doing.