Recover ReplicationGroup from schema change in 1.19+

[chlunde]

Precondition: You have replicationgroups created on version 1.18 or older, and you have upgraded to 1.19 or newer and the provider does not work (crash loop backoff) due to serialization errors.

Relevant issues are:

• [Bug]: atRestEncryptionEnabled type changed to string from boolean #1649
• [Bug]: unmarshal error during upgrade from v1.18.3 to v1.19.0 with same scheme version for ReplicationGroup.elasticache.aws.upbound.io/v1beta2 #1682
• [Bug]: ReplicationGroup.elasticache.aws.upbound.io/v1beta2 fails to sync when transitEncryptionEnabled is false #1654

Step 1: Get control over what you use for atRestEncryptionEnabled today

The following steps assume you use true in all cases, if not, you will have to adjust.

kubectl get replicationgroups.elasticache.aws.upbound.io  -o json |
    jq '.items[].spec.forProvider.atRestEncryptionEnabled' | sort | uniq -c
      3 true

Step 2: Replace .spec.forProvider.atRestEncryptionEnabled as string value using patch

for rg in $(kubectl get replicationgroups.elasticache.aws.upbound.io -o name); do
    kubectl patch  -p '{"spec":{"forProvider":{"atRestEncryptionEnabled":"true"}}}' --type=merge "$rg";
done

Step 3: Patch .status.atProvider.atRestEncryptionEnabled

This is the tricky part, as normally you never adjust the status, and kubectl patch, edit will normally ignore your request. The trick is to use --subresource=status:

for rg in $(kubectl get replicationgroups.elasticache.aws.upbound.io -o name); do
    kubectl patch --subresource=status  -p '{"status":{"atProvider":{"atRestEncryptionEnabled":"true"}}}' --type=merge "$rg";
done