I run seti and cpdn on a fedora biX86 system. Cpdn has a low activity percentage, the it is ususally sharing one processor with seti, another seti instance running continuously in the second proc.
Quite often, the seti instance which is running continuously losses contatct with boinc. Seems thet boinc pauses the two seti from time to time, and of course, restarts only one instance when it pauses cpdn.
In that case, killing boinc stops only the seti instance which is active (or switching with cpdn). I have to kill the second seti instance, which is waiting forever,before restaring boinc. Then everything is ok, until the next lost connection..

I have the same problem, much less often, on biproc macosX boxes.

Anyone or boinc team- has a solution ?

added precision: i use the "leave in memory" option for the paused client.

---