502 "Bad Gateway" errors consistently occurring when hitting "Run" on any function in Kudu

As of Friday morning 10/4/2019, we cannot run a WebJob function manually through the Kudu console. When we click “Run”, It spins for 2 minutes and returns a 502 error.

We’re seeing it with multiple completely isolated production WebJobs (different App Service Plans, storage accounts, regions), none of which have had code changes in a while, as well as a totally trivial test WebJob deployed solely to troubleshoot this issue.

We have googled this extensively, been through many troubleshooting guides, and examined our code and app service health carefully. Nothing seems to apply. But the announcement of .NET Core 3.0 availability on Friday makes us wonder if that’s somehow related.

Steps to reproduce:

1. Create a WebJob project using .NET Framework and the WebJobs SDK.
2. Add a trivial no-op function, marked with [NoAutomaticTrigger] (or any trigger attribute).
3. Deploy to a continuous WebJob.
4. Browse to the WebJob and specific function in Kudu.
5. Click the “Run” button.

[tmenier]

I confirmed the work-around is working for us.

If it helps anyone else, the Microsoft account you likely use to log into the Azure portal won’t work for basic auth. You can find the credentials in the portal by navigating to your WebJob, highlight it, click Properties, and grab the username/password from that blade.

[mathewc]

We’ve found a regression in our recent App Service deployment (not related to .NET Core 3 or WebJobs). The issue affects form POSTs to the SCM/Kudu endpoint using SSO login. ARM and basic auth are not impacted and could be used for a workaround (@ahmelsayed). We’re actively working on a patch for this.