Alrighty here we go. As some of you know my 400W Diazo metal halide lamp is cycling, IDK why. Thought it was maybe meant for increased CWA capacitance to run at 500W, that is probably not the case. I have tried it in vertical and horizontal burning positions. Here are the specs I measured:

When the lamp strikes, the voltage is around 15V and the current is around 3.5A. The voltage slowly climbs until it cycles off. But the voltage it reaches before cycling is only 30V (only a few seconds after switch-on). I am fairly certain these are accurate RMS measurements.

Compared to the >300VOC of the ballast, 30V is a 10:1 ocv to lamp voltage ratio, which makes it very very odd for it to cycle off because i would expect it to do that at 150V or more. The capacitor checked out fine.

Anybody have ideas? At my wits end with this thing 

Could suggest electrode failures? Observing the voltage using an oscilloscope may reveal a bit more - what reignition spike is there,... But be careful, so the ignitor won't kill the oscilloscope, you need a HV input probe for that.

And a question: Aren't these industrial lamps designed for forced air cooling, so operating them in a free air may mean the lamp actually overheated and maybe got damaged?

@Medved
Yeah I guess I could just measure it, I don't have a dedicated HV input probe but I'll set up a "good enough" resistive voltage divider.

Usually the lamps that I have seen that were for fan cooling were all tubular instead of BT, but that is an excellent question, that could be.

What is the way you are measuring arc voltage? It is not trivial, as ignitor voltage will fry most multimeters.

It is quite against the physics for the arc to extinguish abruptly once it is already established and lamp runs up with hot electrodes and discharge voltage is just 30V. Though, arc tube may be especially dirty and re-ignition peaks may present and grow about ballast OCV.

As an interesting example, I have one 400W color MH lamp where one of the electrodes slightly touched badly formed arc tube wall and then definitely bad things happened, as electrode likely reduced some of silica, dissolving some silicon in the tungsten, likely poisoning the electrode, and it also released some oxygen to the arctube. So far, the lamp now fails to do glow-to-arc with a classic choke+superimposed ignitor, and stays with a dim glow discharge just initiated by the ignitor, but surprisingly lights up when on 150W electronic squarewave ballast! It attempts to run-up for a minute or so, but then the ballast brains likely decide the lamp is rectifying and shut the thing down.