Aidan

Missed this response somehow. So is the suggestion to try and override `__connect__` to handle connection errors, and then use `RecvTrailingMetadata` to cause a connection error on a specific status...

We are doing memory profiling on containers running in k8s. We can port forward to localhost and do it, but there is an added complication of most devs being on...

Yeah I ran into that when trying out the `--native` traces. The changes I made to support it seemed quite minor, and I left default behavior the same.

https://github.com/lidatong/dataclasses-json/pull/231 Fixed here @lidatong

Thanks for the quick reply, glad to hear it wasn't my code. I'll subscribe to that bug and see what happens.

+1 to setting this with the helm chart. Updating all of the node groups with custom launch templates is quite the chore to configure this one setting.

@d-nishi what needs to be done to have this considered?

/remove-lifecycle stale

/remove-lifecycle stale

FWIW I've been using this implementation since March without issues.