Eno Compton

Another option would be to follow Istio with its [`/quitquitquit`](https://istio.io/latest/news/releases/1.3.x/announcing-1.3/change-notes/) endpoint.

To be clear, setting `term_timeout` makes the proxy wait so long _after_ it receives a SIGTERM. A workaround proposed elsewhere (not finding the comment at the moment) is to start...

For completeness, here's a [blog post](https://cotton-ori.medium.com/how-to-terminate-a-side-car-container-in-kubernetes-job-2468f435ca99) that describes how to use a PID file to accomplish the same. In plain bash the approach translates to: ``` bash $ /cloud_sql_proxy -instances=my-project:us-central1:my-instance=tcp:3306...

@WilliamHZ did you see the documentation here? https://github.com/GoogleCloudPlatform/cloudsql-proxy/tree/main/examples/k8s-sidecar#setting-up-a-service-account Trying to figure out what we might do differently.

What error are you getting when the Proxy tries to connect? Also, have you tried verifying your setup? Step 9 here shows you how to run a test on a...

That error looks like you're not using the workload identity. If you create a test like so: ``` yaml apiVersion: v1 kind: Pod metadata: name: workload-identity-test namespace: default spec: containers:...

I'm going to close this. If there's something actionable we can do differently here, feel free to re-open with any suggestions.

We're currently working to get [v2](https://github.com/GoogleCloudPlatform/cloudsql-proxy/issues/872) shipped. Once that's done, we'll revisit this and adjust the priority as needed.

Another option would be to use https://github.com/karlkfi/kubexit to support dependency ordering.

Bumping up the priority on this. Right now there's not a good built-in way to do this. For people who didn't see the [StackOverflow post](https://stackoverflow.com/a/69036052/927514), a current workaround looks like...