Deploying the SNMP Prometheus Adapter

Deployment to Kubernetes

In a Kubernetes environment, the SNMP Prometheus Adapter is designed to run as a sidecar container inside the same pod as the SBA Gateway. The Prometheus metrics can be accessed over HTTP using localhost, but the log files must be written to a shared location, such as a volume, where they can be accessed by both processes.

Figure 1 shows Prometheus metrics written to log files in a shared location.

Configuration

The following shows a sample SNMP Kubernetes configuration for installing the SNMP Prometheus Adapter as a sidecar to the SBA Gateway.

apiVersion: v1
kind: Service
metadata:
  name: sba
  labels:
    app: sba
spec:
  ports:
  - port: 9098
    name: sba5g
  clusterIP: None
  selector:
    app: sba
 
 
---
 
 
apiVersion: v1
kind: ConfigMap
metadata:
  name: snmp-adapter-config
  namespace: matrixx
data:
  snm-adpater.yaml: |
    snmp-adapter:
      nms:
        default:
          host: snmpserver
          notificationPort: 162
          timeout: 5s
          retries: 2
      alarms:
        logFiles:
          - fileName: /var/log/mtx/nf.log
            pollingFrequency: 5s
      prometheusMetrics:
        endpoints:
          - url: 'http://localhost:9098/metrics'
            pollingFrequency: 5s
 
 
---
 
apiVersion: apps/v1
kind: Deployment
metadata:
  name: sba5g
spec:
  replicas: 2
  selector:
    matchLabels:
      app: sba
  template:
    metadata:
      labels:
        app: sba
    spec:
      imagePullSecrets:
        - name: registrypullsecret
      containers:
 
 
        # SBA CHF
      - name: sba
        image: r.m0012242008.com:18080/sba-5gc-networkfunctions-chf
        ports:
        - containerPort: 9098
          name: sba5g
        env:
          - name: nrf.registerWithNrf
            value: "true"
          - name: nrf.leaderSidecar.enabled
            value: "true"
          - name: nrf.nrfUris
            value: "http://nrf:9098"
        volumeMounts:
          - mountPath: /var/log/mtx
            name: sba-log
 
 
        # Leader Elect Sidecar
      - name: leadersidecar
        image: r.m0012242008.com:18080/sba-5gc-networkfunctions-leadersidecar
        ports:
          - containerPort: 8080
        env:
          - name: lock.name
            value: sba
          - name: lock.identity
            valueFrom:
              fieldRef:
                fieldPath: metadata.name
 
 
        # SNMP Adapter Sidecar
      - name: snmpadapter
        image: r.m0012242008.com:18080/snmp-adapter
        ports:
          - containerPort: 5770
            protocol: UDP
        env:
          - name: spring_profiles_active
            value: sba
        volumeMounts:
          - mountPath: /var/log/mtx
            name: sba-log
          - mountPath: /opt/mtx/conf
            name: snmp-adapter-config
 
 
      volumes:
 
 
          # Log Directory Shared by the SBA CHF and the SNMP Adapter
        - name: sba-logs
          emptyDir: {}
 
 
          # SNMP Adapter Configuration Mount
        - name: snmp-adapter-config
          configMap:
            name: snmp-adapter-config

Troubleshooting the Network Connections

The following networking limitations may apply:

• SNMP server (NMS) must be able to poll the SNMP agent of the SNMP Prometheus Adapter to retrieve the metrics.
• The SNMP Prometheus Adapter is running in a pod, and these instances are created and destroyed as needed. Therefore, there is no fixed set of IP addresses with which the SNMP server (NMS) can be configured. These SNMP servers cannot be configured using only the service as the host, because this would yield polling metrics data from only one pod. The results could depend on which load balancer rules are configured, such as round robin, and so on.

The following Prometheus limitations may apply:

• Prometheus has an issue similar to the networking limitation described above with no fixed set of IP addresses. This issue was solved using an operator that uses the Kubernetes API to fetch the IP addresses of the pods associated with a service. This operator can query each of the pods' endpoints and aggregate the results. The success of this approach depends upon which SNMP server (NMS) implementation you are using.
• If an NMS is configured, the SNMP Prometheus Adapter will send a cold start notification (trap) upon startup. It is possible the SNMP server (NMS) might be able to use this notification to register new instances, but the viability of this approach depends on whether the vendor has implemented this as a feature.

Deployment to Docker

The SNMP Prometheus Adapter application can be run as a standalone Docker image, as shown in the following example:

docker run -e spring_profiles_active=sba -p 5770:5770/udp -v `pwd`/conf:/opt/mtx/conf -v /var/log/mtx:/var/log/mtx r.m0012242008.com:18080/snmp-adapter:5200-SNAPSHOT

This runs a 5200-SNAPSHOT version of the image with the SBA default configuration loaded. It assumes that you have a snmp-adapter.yaml file containing your file and endpoint configuration in a conf sub directory in the same directory as you are running the command from. See Docker CLI: Run for more information and options.

Configuration

The SBA profile is activated by default and a default configuration file is created in the following directory: /opt/mtx/conf/snmp-adapter.yaml.

Logging

The SNMP Prometheus Adapter application logs are written to the log file in the following directory: /var/log/mtx/snmp-adapter.log. The SNMP MIB files can be found in the following directory: /opt/mtx/data/snmp/mibs.