OK, so my dad has this cheapo Regent 70W HPS floodlight (well, two of them). The one in question lasted about 15 years of occasional use before the ignitor crapped out several months ago. My dad got a new Sylvania lamp which did nothing, so I got a cheap ProLume ignitor on eBay, removed the old ignitor, installed the new one by matching up the wire colors, and it worked. Well, now it's dead again after hardly any use. The light has an external PC which may have failed, but assuming the fixture IS getting power, what do you think went wrong? Can an ignitor fail that quickly? Could I have wired it improperly? Again, it DID work when I tested it after installing it.

Does the ignitor or ballast makes any noise when it is supposed to generate the ignition pulses (powered without lamp)?

I assume the ignitor is of a standard range semiparalel type, so a series LR from the Neutral, capacitor from the Lamp wire and a SIDAC from the Ballast tap wire meeting ion one common internal node.
First check, whether the loop formed bu the ignitor "lamp" wire going to the ballast (either directly or via connection at the lamp socket), ballast tap and back to the ignitor is not too long or mainly has not large cross section area (forms a large loop). This path has to carry rather large current spikes and when there is too high stray inductance picked up by the wiring, the pulse from the ignitor gets attenuated. The wire length in this loop shall not exceed about 0.5m (or 1.5 feet) and these two wires to the ignitor are better to be tied together, so the loop is as small as possible.

In a similar way, the pulse gets attenuated when there is large stray capacitance from the Lamp and ballast tap wires to the Ground and/or Neutral. The usual cause is too long wiring to the lamp. Normal ballast-ignitor sets are designed to handle 2m (6feet) of wiring.
"Tidy" wiring style with the Lamp and Neutral wires tied to go parallel is not that optimal here, it causes way higher capacitance than "lousy" wiring. But this bvecome important if you want to exceed the 2m cable length limit for the cable between ballast and lamp, within the 2m limit it should be handled by the ignitor well even when packed together in a compact cable.

Then check the ballast coil: Put a 150W incandescent instead of the HPS lamp, measure voltages. The lamp has to glow a bit dimmer than normally, the voltages should be at around 60..80V range  across the ballast, about 60..100V across the lamp, the ballast should run way colder than normally. It is quire approximate check, however I think it should be capable to detect an eventual short circuit (ballast voltage drop too low) and open circuit (no current flowing, lamps does not light at all) faults.
When the voltages are OK, check the voltage between the tap for the ignitor connection and the Lamp end, there should be something around 5..10V. If there is such low voltage, try to connect there some small incandescent lamp (rated for at least the voltage you have seen). It should light correspondingly to the voltage you have measured. This should obviously check, whether the connection to the ignitor is not broken...

Are the inner connections accessible for some measurements/tests?
Disconnect the ignitor from the fixture.
Check the resistance of the series LR (from the middle point to "Neutral"), compare with the resistor nominal value (you may measure 10..20% higher due to inductor DC resistance)
Then connect the ignitor by the Lamp (the capacitor branch) and Neutral (the resistor and inductor) wires to some 120V AC supply and measure voltage across the capacitor and resistor. From the ratio calculate the capacitance (C = V(LR)/V(C)/R(LR)/(6.28*60)), compare with the printed nominal capacitance. Tolerance of +/-10% should be OK (normal component tolerance plus some measurement error; such tolerance is not important for the ignitor functionality)
Then connect the SIDAC (beiween the middle point and ballast tap connection) in series with about 20W incandescent. It should glow like on a dimmer with slightly reduced power setting. No 60Hz flicker or so.

The typical faults are open circuit resistor, capacitor either shorted or more likely with capacitance way lower than nominal (way less than 10% or so; even 20..30% of nominal usually suffice for the ignitor to successfully ignite the lamps, there is usually quite a lot of margin).

You said it was a cheap e-bay ignitor...there's your answer. (Try to find an ignitor from Advance or another top tier manufacturer.)

Option 1 is to gut the fixture and wire line voltage to the socket. If you only use it on occasion, install a halogen A19 bulb.
If you're going to use it as a security light, install a 23w or 32w 4100K commercial grade TCP CFL.  (I've done this and it works well, and is also very easy to maintain.)

A halogen A19 or a CFL isn't going to cut it. The lights are quite high up and the intensity from the HPS lamp is necessary in this application. I will just have to get up there and see what's wrong...I guess if the fixture is getting power and a good lamp doesn't work, it must be the ignitor. I don't think the ballast would be dead.

I use advance ignitors for MH and HPS and they're pretty reliable. Just don't buy the advance crapacitors. They're plastic dry film. Go for an amrad like a funkybulb would.

These lights are NPF so no capacitor involved.

I haven't had time to look at this again, but assuming I do need a new ignitor, would this one work?

When I get around to looking at this fixture again, I think I'll just try swapping lamps and ignitors between the two fixtures to narrow down the problem.

According to the description it should work.
However strange is, it is just a ě pin device and mainly for the wattage range from 35W till 400W. I wonder how this distinguish between ignited and not yet ignited lamps 9to know, when to generate the pulses0, when for the 35..150W the ballast OCV is just 120V and the normal arc voltage of the 400W the arc voltage is above 150V...

The basic ignitors distinguish according to the voltage across the lamp, but there one type will work for the 35..150W range (with triggering voltage around 80..100V) and different type for the higher wattages (triggering voltage about 160..200V)...

It's hard to read, but according to the label, there is a jumper that must be installed or removed depending on the lamp wattage.

Turns out it was just the lamp. I really expected more from Sylvania. Here is a picture of the lamp.

Turns out it was just the lamp. I really expected more from Sylvania. Here is a picture of the lamp.

Don't be unrealistic with your expectations.  Every manufacturer will have duds.  Even with great quality control measures.  I deal with a number of HPS fixtures and have put in new bulbs from Philips, GE, and Sylvania that were duds or that only lasted a short amount of time.  Doesn't happen often, but every once in awhile.  Same goes for new fluorescent, CFL, even an LED floodlight.  I can go thru 50 boxes of F32T8s and then come across 2 or 3 boxes within a year time frame that had 1 or more bad tube in them.  Some damaged in shipment, some I don't know what happened.  Perhaps a micro crack/hole somewhere that allowed the gas to vent.