Bug: `Check whether Ansible can connect as <user>` – passes when SSH connection fails about trellis

[swalkinshaw]

Thanks for all the details. I can see why that's confusing but I'm sort of wondering if this was done on purpose... though I wouldn't know the reason. The intention behind that check isn't strictly "can you connect via SSH" but the more nuanced "can the root user connect in order to determine which admin_user Trellis tries to use after".

One goal of this current design might have been to not try and override Ansible's default behaviour for SSH failures. Your solution would be a normal failure, not strictly a connection failure. So if we did that, we might want to write out a nicer custom failure message?

[strarsis]

Yes, more nuanced error messages would be great. When there is an SSH issue (KeePass not opened (for KeeAgent)) this task passes, but then it fails later. As this check happens directly at the start, it would be nice to know immediately if something is wrong.