Blue Green Clusters
For different reasons you might want to create a completely new cluster, this can be for many reasons like:
- Broken cluster
- A test of patching the cluster
- Major breaking change
Thanks to this XKF supports performing blue-green deployment on an entire Kubernetes cluster, this is applicable to both Azure and AWS. These docs are intended for both clouds, the main difference is in the naming.
Today XKS does not support any way of only doing blue-green deployment on a specific environment. If you need to perform blue-green deployment on QA you should do it in dev and prod as well. We think that the risk is too great that you by mistake get drift between the modules used in the different clusters.
Workflow​
We assume that the workloads on the clusters are stateless and can run multiple instances.
- Set up a new cluster in the target environment using Terraform
- Verify that the new cluster is functioning as intended
- You will not be able to verify any ingress
- You will not be able to use AZAD-proxy in the newly created cluster
- Change the TXT DNS records over to the newly created cluster
- Verify that the ingress traffic is migrated to the new cluster and it is working as intended
- Destroy the old cluster using terraform
DNS migration​
You can find a small small script bellow to make the migration of DNS easier. As always use at your own risk and make sure that you understand what the script does.
Our recommendation is that you migrate one DNS record manually and verify that the ingress and the new cluster is working as intended, when you know that you can run the script.
Azure​
ENVIRONMENT="dev"
OLD_OWNER_ID="${ENVIRONMENT}-aks1"
NEW_OWNER_ID="${ENVIRONMENT}-aks2"
RESOURCE_GROUP_NAME="rg-${ENVIRONMENT}-we-aks"
ZONE_NAME="${ENVIRONMENT}.domain.se"
ZONE_RECORDS=$(az network dns record-set txt list -g ${RESOURCE_GROUP_NAME} -z ${ZONE_NAME} | jq -rc '.[]')
ZONE_RECORDS_CSV_ARRAY=( $(jq -rc '. | [.name, (.txtRecords[0].value[0] | @base64)] | join(";")' <<< "${ZONE_RECORDS}") )
for ZONE_RECORD_CSV in "${ZONE_RECORDS_CSV_ARRAY[@]}"; do
ZONE_RECORD_NAME=$(awk -F';' '{print $1}' <<< $ZONE_RECORD_CSV)
OLD_ZONE_TXT_VALUE=$(awk -F';' '{print $2}' <<< $ZONE_RECORD_CSV | base64 -d)
if [[ ${OLD_ZONE_TXT_VALUE} =~ "owner=${OLD_OWNER_ID}" ]]; then
NEW_ZONE_TXT_VALUE=${OLD_ZONE_TXT_VALUE/owner=${OLD_OWNER_ID}/owner=${NEW_OWNER_ID}}
echo Updating external-dns owner of ${ZONE_RECORD_NAME}: ${OLD_OWNER_ID} to ${NEW_OWNER_ID}
az network dns record-set txt add-record --resource-group ${RESOURCE_GROUP_NAME} --zone-name ${ZONE_NAME} --record-set-name ${ZONE_RECORD_NAME} --value "${NEW_ZONE_TXT_VALUE}" 1>/dev/null
az network dns record-set txt remove-record --resource-group ${RESOURCE_GROUP_NAME} --zone-name ${ZONE_NAME} --record-set-name ${ZONE_RECORD_NAME} --value "${OLD_ZONE_TXT_VALUE}" 1>/dev/null
fi
done