Illarion Kovalchuk

I would propose just to change default example to something like this: ``` opts := &Opts{} _, err := flags.Parse(opts) if err != nil { if w, ok := err.(*flags.Error);...

@borud flags prins error message by itself, to the stderr, so that this code does exactly the same thing as the original example intended to do, but more correctly.

@foogod oh, ok, that makes sense. So, it seems that there is no possible way to use go-flags properly then? It's a good case for a pool request then. Please...

@skakee It would be nice if you do so, please list specifics. We are not native English speakers, that's why there's a lot of obvious (not for us :) )...

Looks like a very special use case, and there can be infinite number of such use cases, so it seems to be almost impossible to implement them all, unless we...

I think splitting defaults on protocol basis is overkill for now and will confuse users who expect defaults to be as simple as possible.

but forced disconnecting based on disappearing from discovery makes sense in systems when health-check is provided by external system (the one that returns filtered list of backends)

@AmitBawer yes it is known and it's really simple: in case of failure of some backend, its clients get rebalanced to other backends. Other clients are not rebalanced. If the...

In upcoming release 0.8.0 and in current master branch that is implemented using weight load balancer -- your backend 1 should have priority=0 and backend 2 should have priority=1 (or...

It seems that problem is not in your config, but in your healthcheck_graylog.sh script, it reports that backend is dead.