project-management-only/home-tv-station#7: Kubernetes Manifest

The first implementation looks like this

---

apiVersion: v1
kind: Namespace
metadata:
    name: tv-station

---

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: tvstation
  name: tvstation
  namespace: tv-station
spec:
  selector:
    matchLabels:
      app: tvstation
  template:
    metadata:
      labels:
        app: tvstation
    spec:
      containers:
        - name: tvstation
          image: registry.bentasker.co.uk/project-management-only/home-tv-station:0.1b
          imagePullPolicy: IfNotPresent
          env:
          - name: INFLUXDB_URL
            value: "http://192.168.3.84:8086"
          - name: INFLUXDB_BUCKET
            value: "Systemstats"
          ports:
            - containerPort: 80
              name: http-hlsport
              protocol: TCP
            - containerPort: 1935
              name: rtmp-tvport
              protocol: TCP
          resources:
            requests:
              cpu: 500m
              memory: 750Mi
          volumeMounts:
            - mountPath: /media
              name: media
      volumes:
      - name: media
        nfs:
          server: 192.168.3.233
          path: "/volume1/Series"
          readOnly: true

---


apiVersion: v1
kind: Service
metadata:
  name: tvstation
  namespace: tv-station
spec:
  ports:
    - port: 8082
      protocol: TCP
      targetPort: http-hlsport
      name: "hls"
    - port: 1935
      protocol: TCP
      targetPort: rtmp-tvport
      name: "rtmp"
  externalIPs:
     - 192.168.3.23
     - 192.168.3.22
  selector:
    app: tvstation
  sessionAffinity: None
  type: LoadBalancer


---

The resource allocation is entirely arbitrary - figured I'd let it run a bit and look at actual usage in my dashboards

I've defined both allow and blocklists as a config map:

apiVersion: v1
kind: ConfigMap
metadata:
  name: tv-station-config
  namespace: tv-station  
data:
  allowlist.txt:
  blocklist.txt: |
    Prison_School

Those are then mounted in place (commit 68b20fe23dedc98a9bb23b59f807065f81d25e2d adjusts so we check whether these files have >0 lines in them before enabling the associated mode)

It's worth noting, for avoidance of doubt: if the pod rolls, it's not going to be streaming the same episode when it comes back up.

We could potentially maintain some kind of state so that we can pick back up where we left off, but that's not for the initial release.

Based on a quick look at resources, I'm going with

            requests:
              cpu: 1
              memory: 750Mi
            limits:
              cpu: 2
              memory: 1.5Gi

I think those can actually come down quite a bit, but it's obviously going to vary a bit based on what's being processed/streamed. I'll check back on it tomorrow to see what's been used overnight