Bitnami images no longer available

[lambdasierra]

In 2025, Bitnami changed how it provides Docker images and Helm charts. Details here:

• Upcoming changes to the Bitnami catalog (effective August 28th, 2025) bitnami/charts#35164
• How to prepare for the Bitnami Changes coming soon (blog post, 18 August 2025)

As a result, attempting to pull the Bitnami images fails:

$ docker pull bitnami/redis:7.0.11-debian-11-r12
Error response from daemon: failed to resolve reference "docker.io/bitnami/redis:7.0.11-debian-11-r12": docker.io/bitnami/redis:7.0.11-debian-11-r12: not found

For example, the first 3 lines of chapter-2/kind-load.sh attempt to pull Bitnami images:

platforms-on-k8s/chapter-2/kind-load.sh

Lines 1 to 3 in 86387e1

	docker pull bitnami/redis:7.0.11-debian-11-r12
	docker pull bitnami/postgresql:15.3.0-debian-11-r17
	docker pull bitnami/kafka:3.4.1-debian-11-r0

I am not sure of the “best” way to resolve this issue.

Bitnami has moved older images to the Bitnami Legacy Registry.
I have tested the 3 docker pull commands from kind-load.sh, replacing bitnami/ with bitnamilegacy/, and it works.
The blog post says “this is a temporary solution, and we do not plan to keep this registry around for long.”

Bitnami says it “will continue to offer a limited subset of free, latest-version images intended for development use.”

There may be other consequences from this change, such as the availability of Helm charts, that I have not yet tested.

[savisaluwadana]

Guide: Fixing Bitnami Image Errors on Local k3d Clusters (2025/2026 Fix)

The Issue:

1. Bitnami Changes: Since August 2025, Bitnami moved specific image versions to a "Legacy" registry (bitnamilegacy/), causing standard Helm charts to fail with ImagePullBackOff.
2. k3d Bug: The standard k3d image import command often fails with "content digest not found" or "short read" errors when importing large images like Kafka.

The Solution:
We manually inject the images directly into the cluster's server node, bypassing the network and the import tools entirely.

---

Step 0: Identify Your Cluster Node

First, find the exact name of your cluster's server container. Run this command to list your running k3d nodes:

docker ps --filter "name=k3d-*-server-0" --format "{{.Names}}"

Example output: k3d-dev-server-0 or k3d-mycluster-server-0.

Set this as a variable in your terminal so the commands below work automatically (replace YOUR_NODE_NAME with the output from above):

export CLUSTER_NODE=YOUR_NODE_NAME
# Example: export CLUSTER_NODE=k3d-dev-server-0

---

Step 1: Stop Kubernetes from Pulling Online

Tell Kubernetes to use the local images we are about to inject, rather than trying (and failing) to download them from the internet.

kubectl patch statefulset conference-kafka --type='json' -p='[{"op": "replace", "path": "/spec/template/spec/containers/0/imagePullPolicy", "value": "IfNotPresent"}]'
kubectl patch statefulset conference-redis-master --type='json' -p='[{"op": "replace", "path": "/spec/template/spec/containers/0/imagePullPolicy", "value": "IfNotPresent"}]'
kubectl patch statefulset conference-postgresql --type='json' -p='[{"op": "replace", "path": "/spec/template/spec/containers/0/imagePullPolicy", "value": "IfNotPresent"}]'

---

Step 2: Manually Inject the Images

Run these blocks one by one. This process pulls the image from the Legacy registry, re-tags it to match what the Helm chart expects, and forcefully copies it into the cluster node.

Fix Kafka

# 1. Pull from Legacy & Re-tag
docker pull bitnamilegacy/kafka:3.4.1-debian-11-r0
docker tag bitnamilegacy/kafka:3.4.1-debian-11-r0 docker.io/bitnami/kafka:3.4.1-debian-11-r0

# 2. Save to file & Copy to Node
docker save docker.io/bitnami/kafka:3.4.1-debian-11-r0 -o kafka.tar
docker cp kafka.tar $CLUSTER_NODE:/tmp/kafka.tar

# 3. Import internally using containerd (ctr)
docker exec $CLUSTER_NODE ctr -n k8s.io image import /tmp/kafka.tar

# 4. Cleanup
docker exec $CLUSTER_NODE rm /tmp/kafka.tar
rm kafka.tar

Fix Redis

# 1. Pull from Legacy & Re-tag
docker pull bitnamilegacy/redis:7.0.11-debian-11-r12
docker tag bitnamilegacy/redis:7.0.11-debian-11-r12 docker.io/bitnami/redis:7.0.11-debian-11-r12

# 2. Save to file & Copy to Node
docker save docker.io/bitnami/redis:7.0.11-debian-11-r12 -o redis.tar
docker cp redis.tar $CLUSTER_NODE:/tmp/redis.tar

# 3. Import internally using containerd (ctr)
docker exec $CLUSTER_NODE ctr -n k8s.io image import /tmp/redis.tar

# 4. Cleanup
docker exec $CLUSTER_NODE rm /tmp/redis.tar
rm redis.tar

Fix PostgreSQL

Note: We pull r17 from legacy (which exists) but tag it as r7 (which many older Helm charts specifically ask for).

# 1. Pull from Legacy & Re-tag
docker pull bitnamilegacy/postgresql:15.3.0-debian-11-r17
docker tag bitnamilegacy/postgresql:15.3.0-debian-11-r17 docker.io/bitnami/postgresql:15.3.0-debian-11-r7

# 2. Save to file & Copy to Node
docker save docker.io/bitnami/postgresql:15.3.0-debian-11-r7 -o postgres.tar
docker cp postgres.tar $CLUSTER_NODE:/tmp/postgres.tar

# 3. Import internally using containerd (ctr)
docker exec $CLUSTER_NODE ctr -n k8s.io image import /tmp/postgres.tar

# 4. Cleanup
docker exec $CLUSTER_NODE rm /tmp/postgres.tar
rm postgres.tar

---

Step 3: Restart Infrastructure Pods

Now that the images are inside the cluster, delete the stuck pods. They will restart immediately using the local images.

kubectl delete pod conference-kafka-0 conference-redis-master-0 conference-postgresql-0

Step 4: Wake Up Application Services

The application services (like c4p-service) likely crashed multiple times while waiting for the database. Delete them to force a fresh start now that the infrastructure is ready.

kubectl delete pod -l app=c4p-service

Final Check

Wait 30 seconds and verify all pods are running:

kubectl get pods

[ikwuoz]

An easier approach to fix the errors on local KinD cluster

• edit the affected statefulsets with kubectl edit statefulsets.apps conference-kafka
• in the CLI editor, modify the continer image, replacing bitnami -> bitnamilegacy
• repeat the above steps for both conference-postgres and conference-redis-master
• delete the affected pods with kubectl delete po conference-kafka-0 conference-postgresql-0 conference-redis-master-0
• the pods should restart automatically and be running