Assembly changes detected. Restarting host

Question

My Azure Functions were running fine and all of a sudden I am getting several "Assembly changes detected. Restarting host..." messages that is preventing my functions from completing.

I am not deploying new code so not sure what is triggering the Assembly Change event to fire. I was running on the latest version of the runtime and have since reverted to version 1.0.10947 thinking that maybe the underlying runtime was updated, but I'm still getting that line showing up in the logs.

Update Now that @Alexey has helped me track down what is causing the Assembly changes to be detected. I would like to ask if anyone can tell me WHY an assembly change is being detected even-tough I have not changed/redeployed my application.

Answer 1

After looking in your logs we opened an issue https://github.com/Azure/azure-webjobs-sdk-script/issues/1533#issuecomment-303595960.

Your functions had multiple restores but now issue is gone. Restores could be initiated by changing project.json.

Answer 2

If you are stuck with the multiple

Assembly changes detected. Restarting host

I fixed my issue by deleted the log file in the Kudu services:

https://[FunctionAppName].scm.azurewebsites.net/

and follow on the top menu:

Debug Console >> powerShell

And the file log is :

LogFiles >> Application >> Functions >> function >> [Function name]

You can remove the log file.

Answer 3

my 2c.

I was struggling with this issue for ages and not sure what was causing it. I believe I may have the answer.

Our solution has been toying with consumption plans, but pulled back to full App Service Plans because the initiation times were too long for our rather unique usage patterns.

But 2 of the appsetting params were still in place: WEBSITE_CONTENTSHARE And WEBSITE_CONTENTAZUREFILECONNECTIONSTRING.

per: https://learn.microsoft.com/en-us/azure/azure-functions/functions-app-settings#websitecontentazurefileconnectionstring these are ONLY for consumption plans.

I removed them and... touch wood, the issue seems to be resolved.