They just keep dropping off SSH, killing more than one test per CQ run, failing the CQ run.

All reef DUTs are listed at http://shortn/_IueaCsU931

Looking at the history of one of the DUTs: http://shortn/_zhbR6Uaxrq
Each of the >15 minute repair tasks is a servo reset. So the DUT fell off SSH / ping. Then we restarted it via servo and it came back happy.

No wonder it dies mid-test.
We need to figure out why reef is so unstable.

Or, to save pain for skylab, get rid of reef DUTs and get some healthier board on skylab.