Azure Front Door still experiencing issues

Hi AdamLangley-8486,

Welcome to Microsoft Q&A and thank you for posting your query here!

The partial cache purge behavior you're seeing after the recent Azure Front Door outage is known to happen during the recovery phase. Here are some detailed points and steps based on official Microsoft documentation to help clarify and troubleshoot:

1.Cache Purge Propagation Delay:

After you initiate a cache purge, it can take up to 10 minutes for the purge request to fully propagate across all Azure Front Door Points of Presence (POPs) globally. This means cache invalidation is not instantaneous or atomic across edge locations, leading to some files being refreshed earlier than others.

2.Purge Process and Best Practices

• When purging cache, always specify the exact endpoint, domain, and content paths. Wildcard domain purges are not supported; you must specify subdomains explicitly.
• Cache purging in AFD is case-insensitive and ignores query string variations, but the purge itself applies per path. Thus, purging /scripts/app.js affects all query variations of that file but no other distinct files.
• It's recommended to version your assets via URLs when updating files to avoid reliance on purging for instant cache refresh.

3.Post-Outage Impact:

The recent outage was caused by a configuration error, and Microsoft has been rolling back to a last known good state. Due to layered caching (Azure Front Door edge caches, DNS caches, client caches), full resolution takes time after changes or purges.

How to Purge Cache Correctly:

You can purge cache via the Azure Portal, Azure CLI, or PowerShell:

Azure Portal:

• Go to your Front Door profile overview page.
• Select Purge cache at the top.
• Choose the endpoint and domains/subdomains to purge.
• Enter asset paths individually or use /* to purge entire domains.

Azure CLI Example:

az afd endpoint purge \

--resource-group <your-resource-group> \

--profile-name <your-profile-name> \

--endpoint-name <your-endpoint-name> \

--domains <your-domain> \

--content-paths '/path/to/file.ext'

Replace placeholders with your actual values. This command triggers purge propagation to all POPs but, again, expect up to 10 minutes delay source.

Recommendations

• Always check logs and verify purge status for discrepancies.
• Implement asset versioning to minimize dependence on purges.
• Allow sufficient time (~10 minutes) after purge commands before testing for changes.
• If issues persist beyond 10-15 minutes, consider opening a support ticket with Microsoft to investigate cache state inconsistencies due to the outage.

In summary, AFD cache purge is not atomic and can show partial effects right after purge commands, especially in the wake of a major outage and configuration rollback. Waiting for full propagation and using best practices like asset versioning will provide the most reliable result.

i hope the above response helps, please do let me know if you have any questions on this.

Thanks,

Harish.