Dan Winship

I started trying to address John's PRR comments, but without knowing what approach we are going to take, every answer ends up being "we will do this in the way...

@andrewsykim lol, I had originally proposed PreferLocal as a modification to the existing InternalTrafficPolicy KEP (#3010) but then closed that and replaced it with the separate KEP because, at that...

/lgtm > No parameters are passed to preStop hook handlers and OPENSHIFT_SDN_POD > environment variable is never expanded. Use the existing pod environment > through a shell instead. That sounds...

/triage accepted /lgtm /approve

So, the commits should be arranged so that each commit implements one complete "step" of the refactoring. In particular, each commit should still compile (and work). In this case, the...

I'd say we shoud get rid of proxy mode failover. IIRC, when we first added the iptables proxy, we wanted it to be default, but there were some systems that...

So the discussion is confusing, but three important points: - "Endpoints on Linux nodes" vs "endpoints on Windows nodes" is not relevant. The relevant distinction is "endpoints on _this_ node"...

(following up from the sig net meeting) I said it should behave "_as though_ the traffic goes to the external LB", but it's totally fine to _actually_ send it to...

> Changing the behavior to make it more similar to the implementation on Linux is thus 'merely' an optimization. However, I do think that the reasoning behind the optimization for...

@parul5sahoo There is nothing more to be merged to k/k, but the KEP graduation criteria said we'd enable the feature in some of the periodic scale tests, which has not...