mr-miles

Emitting the logs and metrics would definitely be very useful for tracking and alerting. I'd been interpreting the serf health check as synonymous with "agent is functional" rather than "agent...

Also very interested in this. In future all the services will run in containers probably on linux/EKS/fargate but currently we have windows servers with services hosted in iis or running...

Hi, To summarise, our use case is to get services that are currently tied to windows hosts participating in the service mesh with low effort/no code changes, so we can...

That second diagram (one envoy process per service) is the setup we are going for, since it makes migrating the actual services around and treating them completely independently much easier....

Thanks @david-yu for the quick response - I was not aware of that service-defaults usage. And yes, currently I'm registering the destinations as services. So do the destination addresses have...

@david-yu - now on 1.13 and trying to use those TProxy enhancements to terminating gateway to set up a destination. Is there a full working example anywhere to look at?...

Thanks for the reply. I have it returning the dependencies, and that works nicely - but it doesn't cover this case. Hopefully I can be a bit clearer: - when...

I came across a great detailed write up of the issue from the Istio maintainers which also mentions how it is exacerbated when nodes start up https://docs.google.com/document/d/1SQzrFxtcn3o_79OtJYbSHMPuqZNhR3EaEhbkpBVMXAw/edit#

Not sure why it closed it...

Thanks @curtbushko that's great. I wanted to draw your attention to one detail: I believe the failing init container will cause k8s to restart that container, but not respawn the...