5

I'm trying to configure rolling website updates to an IIS web farm to enable zero downtime deployments. I can't identify a reasonable way to stop traffic being routed from Application Gateway to a VMSS instance.

Any Application Gateway updates seem to take 45 minutes so that isn't really feasible.

If I set a short health probe poll and unhealthy threshold limit for the Application Gateway, and use an app_offline.htm file so IIS responds with an error, users may still receive the app_offline.htm page before the Application Gateway deems the instance unhealthy.

I'd really prefer to tell the Application Gateway to stop serving traffic to instance x, update the instance, then tell Application Gateway to resume routing traffic to the instance.

Am I missing something here? Is there a way to perform a rolling zero-downtime deployment with Application Gateways?

4
  • ..but the Azure Load Balancer that fronts your VMSS instances has TCP and HTTP probing capabilities.. why not use them?
    – evilSnobu
    Oct 4, 2017 at 16:54
  • Hey thanks for the comment. There's an Application Gateway sitting in front of the VMSS, not an Azure Load Balancer. Created specifically this way using an ARM Template for SSL Termination and Web Application Firewall features. You can't (or couldn't) proxy traffic from an Application Gateway to another resource (the VMSS) via an Azure Load Balancer.
    – Rob
    Oct 5, 2017 at 10:42
  • Okay, the App Gateway has health probes. What's wrong with Those? Not taking the instances out of rotation?
    – evilSnobu
    Oct 5, 2017 at 14:21
  • @Rob Can you describe how AG and IIS are configured? Does each IIS site live on separate VMSS and AG or you have a single AG/VMSS and multiple IIS sites?
    – Alex S
    Nov 18, 2019 at 18:56

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.