Thanks to debugging help from @kim, we tracked this down to a regression in how flyctl handles wireguard peers that have been reaped in our backend, but are still active locally.
While we’re working on a fix, here’s a way you can try to fix it for now:
- Stop the fly agent with
fly agent stop - Remove the wireguard peer entry from
~/.fly/config.ymlfor the organization you’re having trouble with, including the org name key. - Start the agent and try
fly ping -a orgname
If this works, deploys should work as well.
2. Stop the