[AVM Module Issue]: ptn/ai-ml/AiFoundry private endpoint deployment fails

[mvdbosch]

Check for previous/existing GitHub issues

• I have checked for previous/existing GitHub issues

Issue Type?

Bug

Module Name

avm/ptn/ai-ml/ai-foundry

(Optional) Module Version

No response

Description

When deploying the bicep AVM ptn/Ai-Ml/AiFoundry with a private endpoint (public internet access: disabled) type of deployment, the deployment gets stuck into status deploying and fails after 2-3 hours. No private eindpoints are created and the AIFoundry resource is in status failed. Existing private DNS zones are provided to the module.

I am unsure if this is related to: #5957

Module deploys fine without private endpoints / public access = enabled.

(Optional) Correlation Id

No response

[microsoft-github-policy-service]

Important

The "Needs: Triage 🔍" label must be removed once the triage process is complete!

Tip

For additional guidance on how to triage this issue/PR, see the BRM Issue Triage documentation.

[avm-organizer]

@mvdbosch, thanks for submitting this issue for the avm/ptn/ai-ml/ai-foundry module!

Important

A member of the @Azure/avm-ptn-aiml-aifoundry-module-owners-bicep team will review it soon!

[sethsteenken]

Hi @mvdbosch - thank you for submitting this issue. I do believe this issue is related to some other issues currently being investigated (#5957, #5994, #6001).

Are you hitting this issue pretty consistently? If so, can you share the bicep of your deployment including any other dependent modules especially networking? That may really help in pinpointing the cause of the problem. Thank you!

[yme-linden-atos]

Hi @sethsteenken, I'm experiencing the same issue as @mvdbosch. I have simply used the waf-aligned test code for AI foundry (https://github.com/Azure/bicep-registry-modules/tree/main/avm/ptn/ai-ml/ai-foundry/tests/e2e/waf-aligned) and adapted it slightly so that it could integrate with our landing zones. This deployment fails regularly with the same error, and I was never able to deploy it in one go (kicking off deployment again a few times does usually solve issue)

Would be happy to share the code with you! Just let me know :)