Paul Ingles

* Golang to 1.15 * Alpine to 3.12 We'll publish as a new beta but that should be the end of anything necessary for v4.

We've noted in other issues (#425 and #217) that the gRPC load-balancing behaviour can get itself into a state where it won't update server addresses. The gRPC team's advice in...

gRPC Resolution

1

This is to help build some tests around the behaviour of gRPC's load balancer and better manage recovery when server addresses don't resolve.

Change localhost references to 127.0.0.1

8

I think it was @roffe that referenced https://pracucci.com/kubernetes-dns-resolution-ndots-options-and-why-it-may-affect-application-performances.html in another issue. I'd suggest we change all the deployment manifest examples to reference `127.0.0.1` rather than `localhost` to avoid DNS during...

Manage TLS with Certificates API

2

Configuring/deploying Kiam currently requires generating TLS certificates to secure credentials in-flight and authenticate the agents and servers. Kubernetes' Certificates API could be used instead: https://kubernetes.io/docs/tasks/tls/managing-tls-in-a-cluster/, potentially simplifying some of the...

Improve TLS error reporting

1

We often get issues created that relate to gRPC TLS problems that aren't immediately obvious from the reported error. For example, in #111: ``` {"level":"warning","msg":"error finding role for pod: rpc...

Bifrost currently doesn't handle rebalancing events- it assumes a single consumer process.

Use prismatic schema to ensure the system config is correct before starting.

Handle manifest validation errors

1

Ensure blueshift handles manifests that fail validation errors. Should log the error and try again after the usual delay. ``` clojure.lang.ExceptionInfo: Invalid manifest. Check map for more details. at clojure.core$ex_info.invoke(core.clj:4403)...

When an import fails query the stl_load_errors table for the errors during that transaction, store them in a file on S3 so its easier to see when/why imports don't process-...