[piped] Please add option to set resource requests and limits, and also add default ones

[aleksasiriski]

Helm chart name

piped

Describe the solution you'd like

Add option to set resource requests and limits

In addition to that set default requests and limits:

frontend:
      resources:
        requests:
          cpu: 32m
          memory: 32Mi
        limits:
          cpu: 128m
          memory: 128Mi
backend:
      resources:
        requests:
          cpu: 500m
          memory: 500Mi
        limits:
          cpu: 1500m
          memory: 1500Mi
ytproxy:
      resources:
        requests:
          cpu: 32m
          memory: 32Mi
        limits:
          cpu: 500m
          memory: 500Mi

Additional Information

No response

[samip5]

I have no actual idea on what the defaults should be and that's why there aren't any set by default.

[aleksasiriski]

But how do I set my own limits?

Also the above mentioned limits were set as a result of inspecting usage when watching a video with kubectl top

[samip5]

But how do I set my own limits?

Like you did previously and upgrading to the newer chart, eg the resources definition is right, but it wasn't exposed before.

[FireMasterK]

I'll suggest some defaults: Backend: max - 1GB to 1.5GB (the heap size is limited to 1GB, however the RSS memory of the jvm can be higher) Proxy: max 300-500MB min 32MB (it can use a lot of ram if there's a lot of users) Frontend: max 128MB min 32MB

[aleksasiriski]

I've edited the config above with suggested changes.

[aleksasiriski]

@samip5 It's not working with this config with chart 2.0.2:

frontend:
  replicas: 2
  resources:
    requests:
      cpu: 32m
      memory: 32Mi
    limits:
      cpu: 128m
      memory: 128Mi
  env:
    BACKEND_HOSTNAME: api.piped.mydomain.com
backend:
  replicas: 2
  resources:
    requests:
      cpu: 500m
      memory: 500Mi
    limits:
      cpu: 1500m
      memory: 1500Mi
  config:
    PORT: 8080
    NUM_WORKERS: 2
    PROXY_PART: https://proxy.piped.mydomain.com
    API_URL: https://api.piped.mydomain.com
    FRONTEND_URL: https://piped.mydomain.com
    COMPROMISED_PASSWORD_CHECK: true
    DISABLE_REGISTRATION: true
    database:
      connection_url: jdbc:postgresql://cloudnative-pg-rw.cnpg-system.svc.cluster.local:5432/piped
      driver_class: org.postgresql.Driver
      dialect: org.hibernate.dialect.PostgreSQLDialect
      username: piped
      password: mypipeddbpassword
ytproxy:
  replicas: 2
  resources:
    requests:
      cpu: 32m
      memory: 32Mi
    limits:
      cpu: 500m
      memory: 500Mi
ingress:
  main:
    enabled: false
  backend:
    enabled: false
  ytproxy:
    enabled: false
postgresql:
  enabled: false

replicas works without issue, but requests and limits are nowhere to be found with kubectl describe

Containers:
  piped-backend:
    Container ID:   containerd://cxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
    Image:          1337kavin/piped:latest
    Image ID:       docker.io/1337kavin/piped@sha256:60802ef9685281955015d7c301ff87b5d2163657bbbf94353bcbbf6d784226ee
    Port:           8080/TCP
    Host Port:      0/TCP
    State:          Running
      Started:      Tue, 18 Apr 2023 11:54:32 +0200
    Ready:          True
    Restart Count:  0
    Liveness:       tcp-socket :8080 delay=0s timeout=1s period=10s #success=1 #failure=3
    Readiness:      tcp-socket :8080 delay=0s timeout=1s period=10s #success=1 #failure=3
    Startup:        tcp-socket :8080 delay=0s timeout=1s period=5s #success=1 #failure=30
    Environment:    <none>
    Mounts:
      /app/config.properties from config-volume (ro,path="config.properties")
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-z5jkk (ro)

[aleksasiriski]

Also, there's no need to close issues / feature requests that aren't 100% resolved.

[aleksasiriski]

Are requests perhaps missing in the deployment? I see them added only to pod template file... Maybe my assumption is wrong, I've just started learning Helm

[samip5]

Maybe this helps: https://github.com/TeamPiped/Piped-Kubernetes/commit/1bb362f205bddd7ae77304c7a7e2549d175aa277

[aleksasiriski]

This didn't fix it, still same config and same output from kubectl describe

[aleksasiriski]

I'll look into making a PR.

[aleksasiriski]

I was busy... Fixed it now, it was wrongly put in Pod definition, but resources are set for individual containers. Also stumbled upon a bug for ytproxy settings.