Open anvddriesch opened 1 year ago
opsctl create etcdbackup
doesn't work for CAPI clusters because of this reason.
➜ opsctl create etcdbackup -i guppy --workloadcluster etcd1
Error: Invalid config: k8srestconfig.Config.Address must not be empty when not using k8srestconfig.Config.InCluster or k8srestconfig.Config.KubeConfig
/home/runner/go/pkg/mod/github.com/giantswarm/k8sclient/v7@v7.0.1/pkg/k8srestconfig/rest_config.go:108
/home/runner/work/opsctl/opsctl/pkg/k8sclient/k8sclient.go:158
/home/runner/work/opsctl/opsctl/command/create/etcdbackup/command.go:201
/home/runner/work/opsctl/opsctl/command/create/etcdbackup/command.go:124
/home/runner/work/opsctl/opsctl/main.go:57
To increase verbosity, re-run with --level=debug
and we don't have --kubeconfig
flag for this command, which is the workaround for opsctl deploy
command. Should I create a separate issue for this?
We talked about this in the team and think we can work on it after current projects, ~ in a few weeks.
opsctl
(different authentication methods for automation and human access)opsctl
revisit this after cert management for capi is done (potentially including vault)
@anvddriesch do you believe this is relevant enough that someone would work on it anytime soon? otherwise I#d say we can close it
Related https://github.com/giantswarm/roadmap/issues/1577
Kubeconfig management in
opsctl
has grown over time and by now is quite convoluted and hard to follow. Furthermore, with CAPI installations not running vault and client certificate creation therefore not being possible, some commands fail. Let's have a thorough look at the current state, do a refactor and fix the broken commands.