Timeout settings being ignored #226
Description
We are experiencing a couple of issues around timeouts. We are deploying Windows workloads via a Deployment manifest, which can take a significant amount of time to fully rollout and we have been trying to extend these timeouts without success.
For example, we have tried using the timeout block as follows
resource "kubectl_manifest" "my_deployment" {
timeouts {
create = "30m"
}
yaml_body = <<YAML
apiVersion: apps/v1
kind: Deployment
...
YAML
}Results:
...kubectl_manifest.my_deployment: Still modifying... [id=/apis/apps/v1/namespaces/my_namespaces/deployments/my_deployment, 20m0s elapsed]
│ Error: my_deployment failed to rollout Deployment
We were expecting the timeout to be extended to 30 minutes, but the apply consistently fails after 20 minutes, which appears to be a default timeout.
In the case described above, the rollout had completed around the 13 minute mark but kept reporting "Still modifying..." until it errored out after 20 minutes.
Furthermore, we have noticed that anything taking longer than 10 minutes hangs waiting for the rollout to complete until it times out after 20 minutes. If the rollout completes after 10 minutes, the provider keeps reporting "Still modifying..." even though the rollout has completed. If the rollout completes in under 10 minutes, everything works as expected.
This appears to be very similar to the issue reported here: #202
Provider Version: 2.1.3
Kubernetes Version : 1.31
Terraform Version: 1.9.6
Any suggestions on how to extend these timeouts would be greatly appreciated...
Also, thank you for maintaining this provider. It has allowed us to work around many issues related to the Hashicorp provider not supporting some of the quirks encountered when running Windows workloads. Being able to Terraform raw manifest files has allowed us to get around many of these limitations.