Unhandled exception when writing to toggle file during startup about unleash-client-dotnet HOT 6 CLOSED

@daveleek FYI, this happens in a Docker with no volumes mounted. There's only 1 process in that container, and that's the ASP.NET Core app using Unleash.

from unleash-client-dotnet.

Thank you for bringing this up @SMLVE. If you don't mind, what's your polling interval set to (if not default)?

from unleash-client-dotnet.

It's set to 30 seconds, which is the default I think.

from unleash-client-dotnet.

Thanks @SMLVE!

So we have identified a likely cause for this. When an instance tag is not provided to unleash upon creation it will create an instance tag with a random number in it for uniqueness. The instance tag makes its way into the name of the toggles backup file.

The random number has a low entropy so collisions could be occasionally occurring when more than one instance of an application running Unleash .NET SDK runs on the same machine/has the same temp folder volume mapped (which could explain why you're unable to reproduce locally and only see it sporadically in production)

Unleash catches and logs an error+exception for this (so unhandled is perhaps poorly worded in the error message), and continues operation.

We're changing the logging for this to log a warning+exception, and setting fallback instance tag to include a GUID component instead of a low entropy random number. We'll release this sometime this week!

// David

from unleash-client-dotnet.

Thanks, for the detailed explanation and for getting back to us so quickly. We are going to update our package whenever the fix is released!

from unleash-client-dotnet.

FYI: Closing as we released an improvement on randomization of instance tags in v 3.0.0

from unleash-client-dotnet.