Hi, we’re configuring persistent volume for the postgres database but having issues connecting to it. What do we need to set in the helm chart?

Will ask the team about this.

The helm chart failed with the message ‘timed out waiting for condition’ and the container in AKS says ‘Error: secret “prefect-server-postgresql” not found’

Hello Scarlett! Can you confirm that there are no secrets in the namespace with prefect server (

kubectl get secret

)? And did you set

existingSecret

in the helm chart?

Hi, we don’t have secret set up yet. We’re using the postgresqlPassword to test in the mean time

I have tested option with setting only postgresqlPassword, it works as intended, so I guess the issue might be with retrieving the value from KeyVault or passing it to the helm chart. I would recommend to create a kubernetes secret with key

postgresql-password

and value from your KeyVault, and set existingSecret to the name of the Kubernetes secret you've created, then it should work.

Hi @Mariia Kerimova, we’re still getting error. Is it possible that I send over an email on what we have tried and the errors we’re getting?

Sure

A colleague and I went through a similar process today. Hints in this thread helped us get over to finally having a running server. Of course we would likely still be scratching our heads if it wasn't for the helm charts and the . It's a bit of a shame that Laura didn't actually run the upgrade for using external postgres instance, it may have clarified some things. Steps we stumbled on (we're kubernetes/helm newbies): • Like @Scarlett King, we observed that we couldn't get things running by using

postgresql.postgresqlPassword

with

postgresql.existingSecret

set to

null

. Did anyone get this running? • When using

postgresql.existingSecret

we misread the comment and thought that the secret should be called

postgresql-password

, rather than the key. We should have paid more attention. • It looked like the setup expects a database called

prefect

to exist in postgresql instance. Once we created it manually things progressed further. The init container seems to assume it exists and starts with upgrade. • Last piece of the puzzle was that our postgresql instance was at quite an old version (9.6) and didn't support all the syntax in migration scripts - once we upgraded postgresql to 11 (Spinal Tap reference coincidental) we were finally able to get the whole thing running. Hope this helps, either to folks going down the same route or you Prefect folks with updating docs/scripts. And finally big shout out to @Laura Lorenz ❤️ for the youtube presentation - it made things look a lot less scary and @Zanie ❤️ for getting the helm chart together - without it we may not even embark on this journey or may have abandoned it at some point.

Hello Dali! Thank you very much for your feedback! I'll open a PR to update the Readme of the helm chart about Postgresql version. 🙏

Glad you got this working! Mariia has opened a PR to the helm chart docs but feel free to PR any other documentation/faqs that you think may be helpful.