Long time customer shipped a 45 SET amp in for repair, it was nuking output tubes.
It was a mess internally, at some time, someone replaced the coupling caps with giant Russian Teflon caps and didn't actually secure them, just held on with a sticky pad. They flew around in the amp, in their travels, they ripped the ground off one of the input cables. I fashioned some clamps from AL sheet, quick and dirty, not beautiful, but they are inside and sufficient for the task.
All the holes in the chassis for the transformer wiring had not been de-burred, sharp edges abounded. My customer didn't want to invest the time in having me remove all the iron and clean up the holes. I opted for a little hot melt glue around the edges, makes it far less likely the insulation will be compromised. Not ideal, but 5 minutes vs a couple hours. I also swapped the wires to the fuse holder, this way the outer edge isn't hot when the amp is plugged in, removing the risk of shock if replacing the fuse while connected to the mains. I saw it happen to a friend, decided to correct that on anything I worked on. Easy and I sleep better.
The root cause was too much B+, the RCA data sheet for the 45 suggests 250V on the plate, with 34mA of bias current and a max of 275V. This had a bit over 350, even with ~50V across the cathode resistor, that was 300V across the tube. After a while, one would get too hot, draw even more current and run away, incinerating it in the process. I added a 500ohm/25W power resistor between the rectifier and first filter cap and chokes.. It's dropping about 35V and now it has 305V on the plates and with the ~48V on the cathode resistor, we're right in the ball park, about 255-260V across the tube. Hopefully no more problems for the owner or future owner. It can manage 1.7W at clipping, but of course that is about 5% THD, it is, after all, a single ended amp with no feedback.
And yes, my bench is messy, I clean it every now and then... but it quickly reverts to a mess. :)