Provide ARM support for Backstage Showcase with PostgreSQL

[Bill Maniates]

When I attempt to deploy this chart, the backstage application pod fails to run. I tried the latest image, as well as some of the older image tags as well.

$ oc logs -f po/backstage-5cf5cb6ccd-ppwcv
exec /usr/bin/node: exec format error

$ oc describe deployment/backstage
Name: backstage
Namespace: default
CreationTimestamp: Mon, 30 Jan 2023 10:23:16 -0600
Labels: app.kubernetes.io/instance=backstage
app.kubernetes.io/managed-by=Helm
app.kubernetes.io/name=backstage
app.kubernetes.io/version=v1.7.0
helm.sh/chart=backstage-0.1.6
Annotations: deployment.kubernetes.io/revision: 7
meta.helm.sh/release-name: backstage
meta.helm.sh/release-namespace: default
Selector: app.kubernetes.io/instance=backstage,app.kubernetes.io/name=backstage
Replicas: 1 desired | 1 updated | 1 total | 0 available | 1 unavailable
StrategyType: RollingUpdate
MinReadySeconds: 0
RollingUpdateStrategy: 25% max unavailable, 25% max surge
Pod Template:
Labels: app.kubernetes.io/instance=backstage
app.kubernetes.io/name=backstage
Service Account: backstage
Containers:
backstage:
Image: ghcr.io/janus-idp/redhat-backstage-build:500312b32dc375b5326550c498aadeeeac1b33f9
Port: 8080/TCP
Host Port: 0/TCP
Command:
node
packages/backend
--config
/config/app-config.yaml
Limits:
cpu: 100m
memory: 128Mi
Requests:
cpu: 100m
memory: 128Mi
Liveness: http-get http://:http/ delay=0s timeout=1s period=10s #success=1 #failure=3
Readiness: http-get http://:http/ delay=0s timeout=1s period=10s #success=1 #failure=3
Environment:
POSTGRES_ADMIN_PASSWORD: <set to the key 'databasePassword' in secret 'backstage-postgresql'> Optional: false
POSTGRES_HOST: backstage-postgresql.default.svc
POSTGRES_USER: postgres
POSTGRES_PORT: 5432
Mounts:
/config from backstage-config (ro)
Volumes:
backstage-config:
Type: Secret (a volume populated by a Secret)
SecretName: backstage-config
Optional: false
Conditions:
Type Status Reason

---

Progressing True NewReplicaSetAvailable
Available False MinimumReplicasUnavailable
OldReplicaSets:
NewReplicaSet: backstage-5cf5cb6ccd (1/1 replicas created)
Events:
Type Reason Age From Message

---

Normal ScalingReplicaSet 39m deployment-controller Scaled down replica set backstage-5cf5cb6ccd to 0
Normal ScalingReplicaSet 2m6s (x2 over 2d18h) deployment-controller Scaled up replica set backstage-5cf5cb6ccd to 1

[James Blair]

Hey @billmania42 Exec format error can occur if your computer architecture is different than what the image was built for. Can you please confirm what cpu architecture you are using?

The image specified in the values.yaml for this chart is https://quay.io/repository/janus-idp/redhat-backstage-build?tab=tags&tag=latest

That image currently only supports arm64 and amd64 cpu architectures.

[Monica Granbois]

I get the same error.

I am attempting to run locally with Red Hat OpenShift Local on my Mac, which uses the Apple M1 Pro chip. It uses the arm64 architecture.

The log file in the postgres pod gives the following error:

exec /usr/bin/container-entrypoint: exec format error

[Tom Coufal]

Hi @MonicaG , what container image are you deploying? Also redhat-backstage-build or backstage-showcase? Why am I asking.. this chart very recently switched the default image from redhat-backstage-build to backstage-showcase and that may be the source of your problems since this image is x86_64 only at this moment.

In any case we need to look into this 🙂

[Monica Granbois]

Hi @tumido - The janus-idp/backstage-showcase image. I copied the values.yaml file and updated the apps.example.com to .apps-crc.testing. Other than that change, the file was the same.

I tried with the redhat-backstage-build image as well. I received the same error.

Could it be something with the fedora/postgresql-15 image? It is the postgresql pod that is displaying the error in the log.

[Tom Coufal]

I'll try it on OCP local on M1 and get back to you.🙂 Sorry for the inconvenience.

[Tom Coufal]

@MonicaG I've created this experimental ARM image can you please try it?

And you are also right that the fedora/postgresql-15 is not ARM compatible either. Fortunately, there's an alternative already available within OpenShift Local.

Can you please try following values:

helm install oci://ghcr.io/janus-idp/helm-backstage/backstage \
  --generate-name \
  --set upstream.backstage.image.repository=tcoufal/backstage-showcase \
  --set upstream.backstage.image.tag=arm \
  --set upstream.postgresql.image.registry=default-route-openshift-image-registry.apps-crc.testing \
  --set upstream.postgresql.image.repository=openshift/postgresql

The command above doesn't address proper hostname configuration, but it should at least properly deploy. At least it does on my machine. I hope we can simplify the hostname configuration via #63 🤷

[Monica Granbois]

Hi @tumido - yes, it works on my computer! Thank you for this fix!