Improvement
Currently, the Operator only checks the health of the newly created pod and then reports the deployment as successful.
In some scenarios, the pod can crash after and be stuck in crashloop state, blocking the rollout of the deployment.
In this case, the deployment should be considered as failed and be reported like that.
Reason
In the current way, we send inaccurate information, people imagine their deployment is successful when in reality it never went through, and the old version is still present and serving the traffic.
Information
The Operator should look at the state of the deployment and wait for the rollout to be successful before sending the final state DEPLOY_STORE_SUCCESS
Definition of Done
The Operator waiting the rollout is successful before sending DEPLOY_STORE_SUCCESS for all 3 deployment (storefront, admin, worker)
Improvement
Currently, the Operator only checks the health of the newly created pod and then reports the deployment as successful.
In some scenarios, the pod can crash after and be stuck in
crashloopstate, blocking the rollout of the deployment.In this case, the deployment should be considered as
failedand be reported like that.Reason
In the current way, we send inaccurate information, people imagine their deployment is successful when in reality it never went through, and the old version is still present and serving the traffic.
Information
The Operator should look at the state of the deployment and wait for the rollout to be successful before sending the final state
DEPLOY_STORE_SUCCESSDefinition of Done
The Operator waiting the rollout is successful before sending
DEPLOY_STORE_SUCCESSfor all 3 deployment (storefront, admin, worker)