siderolabs / talos

Talos Linux is a modern Linux distribution built for Kubernetes.
https://www.talos.dev
Mozilla Public License 2.0
5.72k stars 464 forks source link

Talosctl option for troubleshooting #7255

Open tba77 opened 1 year ago

tba77 commented 1 year ago

Sometimes we have some issues with our talos cluster which is not related to any miss-configuration or mistake but rather due to some issues in other services like dns, network etc... So I would suggest to add some troubleshooting commands to talosctl to be able to ping other servers, query dns or check a tcp port. It would be great to have commands like

talosctl ping <ip address>
talosctl trace(route) <ip address>
talosctl nc <ip_address> <tcp_port>
talosctl dig @dns-server domaine_name type
netthier commented 1 year ago

Not what you proposed but talosctl has a pcap subcommand which could prove useful for troubleshooting. And if the cluster isn't too broken, it should be possible to generate the appropriate ICMP/DNS/TCP traffic from within pods.

tba77 commented 1 year ago

Yes we can do that with pods also, but it thought it could be interesting to have some useful diagnostic commands, as you said sometimes we can't create pods for that purpose.