There were issues in the past with endpoints losing their connection.
It can be many different things.
If the gateway that the endpoint is associated with goes down then the endpoint is "orphaned" for a time. It will not try to come back into the TMR until a pre-configured time (entry in last.cfg and lcf.dat files).
Quick solution to try is when your endpoint tries to startup again it will use the lcf.dat file if it is present. This is an "encrypted" form of last.cfg, so don't edit it, but if you remove it then it should fall back to the last.,cfg file which you can edit.
Let me know how this goes and what version you're on (also what version are your endpoints running) and I'll see if I can help further.