Open
Description
We have a kind of primitive stab at a method to clean up a faulty runspace state in the PSES pipeline thread consumer here:
We need to induce a fault and see if this works as expected, as well as ensure it doesn't run at other times...
My suspicion is that it will need some refinement and could have some significant bugs in it. But it's currently a corner case and a worse-case scenario handler anyway (if it didn't exist, we are essentially crashed).
So we need to investigate this and see how we can improve:
- detection of bad runspace state
- restoration of good runspace state
- how our threading mechanisms work in this scenario...