Hello flyte-squadron, I'm attempting to deploy

flyte-core v1.10.6

via terragrunt on gcp and the apply is failing due to the helm release for flyte timing out (context deadline exceeded). I've identified that the init run-migrations container for

flyteadmin

and

datacatalog

is failing to connect to the database: `failed to initialize database, got error failed to connect to `host=DB_IP user=DB_USER database=flyte`: dial error (timeout: dial tcp DB_IP5432 connect: connection timed out)` Do you all have any suggestions for triaging this failing database connection? I've copied the flyteadmin run-migrations container spec, and have modified it to run ubuntu to allow me to test with bash. With that manual spec applied (with all secrets mounted), I was able to reproduce connection timeout with

psql

but wasn't able to find any more clues towards network misconfigurations. *Now thinking it has to be a vpc firewall misconfiguration. So less of a flyte question but any net test tips greatly appreciated

you did the right thing, psql needs to work for flyte to work. usually on AWS this would mean security groups. on gcp a gcp expert here can point it

Either cloudsqlproxy (have tried this, but didn't get it to work) or have your cluster/VM and postgres DB in the same VPC (see https://github.com/unionai-oss/deploy-flyte/blob/main/environments/gcp/flyte-core/README.md for an example)

Thank you all very much for the references! To clarify, this is a GCP based deployment. Cloudsql proxy approach is preferred by us but after an initial attempt at that, I wanted to start with direct connection with everything on the same VPC. There is likely something misconfigured VPC/Firewall rule wise, Trying to get the gcloud network connectivity tests to shed some light on what wrong there.

If you make progress on the cloudsql proxy approach, would be glad to hear about it!