I realized I didn't put this in the log on Friday.
I pulled the beamline (hacweb5) crate from the hvgui_hrs config for now (not working anyway). I'm not sure what is in there, but a lot of HVs are on -- nobody has been complaining about the GUI being broken, so I'll assume that 'on' is the desired state.
Work in progress:
I put a line booster on the serial connection (thanks Jack), but the java gui is still erroring out at the same place. When I dig deeper I still can't see any glitches on the hacweb5 end -- no errors are thrown there and I can talk to the crate 'manually' without any problems. Bit of a puzzle unless one of the cards is sending something stupid...
The java gui appears to be dying when it attempts to build the graphics interface for that crate -- presumbably it is getting something strange from the HV server, but I can't see what it is choking on yet. The backtrace that java is spewing is not very helpful either...
Anyway, I was going to poke at the java code this weekend from home.