We’ve started the Auto-Purge work, with this current design: https://github.com/AzureCR/specs/tree/master/auto-purge
Please provide feedback here: https://github.com/AzureCR/specs/issues/1
While we haven't finished the full auto-purge policy, we have made some progress to make things easier:
We’ve started some work in Helm 3 to enable cleaner integration with Helm and Registries. https://github.com/helm/community/pull/55
This work is underway, within the OCI image and distribution sepcs, and Helm 3 repository work.
Thanks for the push.
Apologies. We always balance performance and requirements. In this case, we slightly over pivoted on performance to regenerate the ever growing index.yaml. We’ve rolled out changes to remove the gap between a Helm Chart push and repo update. We’re also working with the community on making Helm Chart registries a mainstream scenario.
If you still see latency, please open an issue at https://AKS.ms/ACR/issues
We completely agree with the scenario you're shooting for, where ACR Notifications should be resilient and idempotent. We looked at making Webhooks more resilient, but also saw the pattern where customers want events from multiple services. Rather than each service build up this same capability, we've started onboarding to Event Grid, which provides these core capabilities. As teams consume events from Event Grid, they get a common integration point.
However, reading the example, I see you're not really asking for your consumption of Webhooks, rather App Service to be the consumer here.
This is something we've discussed, and I'll renew the conversation. Thank you for the post, and if I've missunderstood the example, please ping back.