Make error "Weave is not running" less confusing when running on Kubernetes

[brb]

We suggest to run weave reset when weaver detects an existing bridge of different type than the requested one.

In case of Kubernetes, if a user runs weave reset, it returns Weave is not running which might make the user think that weave reset was not successful (happened in #3318).

[bboreham]

Whilst this change may help, it isn't that hard to detect that it is running.
Also the current instructions to run weave status in Kubernetes are very complicated. Maybe we should provide a container which "does the right thing" and can be run by kubectl run ?

[brb]

Totally agree. However, creating such container will take us longer than changing this error message. So, as a temporary solution I suggested the change to the message.

[brb]

Created an issue for your suggestion: #3321

[brb]

IMO, in case of k8s, we should improve the error message Existing bridge type "bridge" is different than requested "bridged_fastdp". Please do 'weave reset' and try again too, as it does not say whether a user needs to delete the weave-kube deployment and only afterwards to run the cmd.

[hswong3i]

Could we make it commit so a 2.4.0 release?

[bboreham]

@hswong3i I can't parse your comment. Are you asking if you can submit a PR to solve this issue? That would be most welcome.

[hswong3i]

@bboreham sorry for my poor english, I am looking forward for a v2.4.0 release, so I could include it into kubespray, e.g. kubernetes-sigs/kubespray#2991. Hopefully we could tidy up the remaining issue for v2.4.0?

[bboreham]

@hswong3i ok, we plan to get #3313 into 2.4 which is a big change so taking a while to review. Hopefully not too much longer.