Bug Fixes
-
lager ssh-setupandlager updateno longer install their own key on a box a control plane manages. That control plane installs the key of every user it grants, so a key added here is one it does not know about and cannot revoke. Both commands now use the key that is already on the box, and name the one they used. On a box that no control plane manages, nothing changes: the commands generate, install and register a key as before. -
A box you have no access to asks for a grant, not a password.
lager ssh-setupagainst such a box tells you to ask an admin for access in the control plane, which installs your key on every box you are granted and removes it again when you are not. A key installed to reach the box is taken back out, and a removal that fails says so. -
A box that refuses passwords no longer reports a wrong password. Both commands ask the
box’s SSH server which authentication methods it accepts before they offer to install a key.
A box with password authentication off now says that, instead of running
ssh-copy-idand blaming the password you typed. -
lager debug SWD flashandlager debug SWD erasework again on a DA1469x. The RAM-resident flash loader was entered with the Thumb bit clear, so the core faulted on its first instruction and locked up. The loader is now entered in Thumb state, and a core that locks up on entry says so instead of being retried. The missing-loader message also names a command that exists:lager ssh --box <box>. -
lager dac,lager adc,lager gpiandlager gponame the supported instruments when they refuse a net’s instrument. A bareLabJack, which names no model, and aLabJack_T4orLabJack_T8, which no driver supports, now get an error that lists the models the net can use. -
lager boxes unlockreleases a lock that the CLI already counts as yours, without--force. This includes the auto-lock that an earlier step of the same CI job left, and a lock that another tool took under your email. The new--useroption names the holder of a lock recorded under a different name. -
Ending a
lager uartsession no longer logs a read error on the box. A read that fails because teardown closed the port is not reported, and a session thatlager uart --forcedisplaces is told that another client took the net.

