Have this new 70w CMH bulb which has difficulties starting on its intended ballast (M98), sometimes it starts after a while (<1 min), sometimes not. When it starts, the lamp runs normally, same issue both on an electronic and magnetic ballast. Double checked the wiring and supply voltage, all normal. Other 70w MH bulbs work fine on those ballasts. Any ideas to what might be causing this?

I had once problem with a Sylvania (SLI - it is in Europe) 70W Rx7s "not starting", the culprit I found as the cement supporting the small contact disc on the socket had disintegrated and the disc hat tilted, leading to an intermittent contact in the socket...

I have a couple of CMH fixtures running electronic ballasts and sometimes, out of the blue, they take 1-2 minutes to strike. Usually they take less than 5 seconds to strike. I have always thought it was sort of normal for it to happen once in a while, just like pink/orange flashes can be sometimes intense, sometimes inexistant.

Appears operating the bulb horizontally (as opposed to base down before) results in consistent <1 minute startups, even started right away when I tried it yesterday. Still confused what's going on here, since the bulb is rated for universal operation so I might eventually try base up runs out of curiosity as well.

My workplace's parking lot is still lit with MH shoeboxes, at least for the time being (well, minus one pole that got knocked down by the snow plow guy last winter, but I digress    ) And I've noticed that not all the lamps ignite immediately. Some do but a few need as much as 10 seconds to start, and they don't even appear close to EOL yet.

Maybe is due replacement of Kr85 with other starting methods?

This lamp does cause issues with electronic ballasts which have a short time out period (~20s), but is fine on magnetic gear though

20 seconds is unusually short for ignitor time out. I have some that have a temporary timeout at 20 seconds, but it retries about once every 30 seconds ? until about 20 minutes when it shuts down for good, until power is cycled.

I've seen how Eltam ES-PI 1000 ignitor, behaves with a failed lamp: It tries 5 secs every 1 min, three times before giving up.
My Philips HID-PrimaVision PV C 70 /S CDM electronic ballast in my tracklight, don't have timed pulsing, and it isn't a digital ballast, but just have a timer to stop the operation of the ignitor after 15 min of unsuccessful ignition.

But definitely needing to pulse for longer than just few seconds means there is something wrong with the lamp, ballast or the combination (that particular lamp may require higher pulse voltage,...)
Question: Are the ignition attempts coming with sparks within the arctube? If not, the reason could be too low ignitor voltage for that lamp. Or there is electrical breakdown somewhere (the base, socket, could be contaminated atmosphere in the outer,...).

@Medved to answer your questions - no, there are no arc tube flashes, nor anywhere in the bulb. Screw base appears to have good contact too, perhaps an issue with the fill gas?

Also might be worth noting this lamp doesn't contain either Kr85 (no notice on the package) nor a UV glow enhancer capsule.

So it seems the ignitor peak voltage is below what the lamp needs. So either the ignition pulses get reduced by something, or the lamp needs higher voltage than it is supposed to.

Dunno if that is your case, but too long wiring between the lamp and ballast could reduce the ignition pulse. Normally standard ignitors (include those in the electronic ballasts) are rated for maximum 2m length, but better to stay within 50cm limit.
Or broken insulator glass of the center contact in the lamp base may trap some dirt and humidity, so cause flashovers there.

Or the arctuba is just contaminated...

For the Kr85: I've rather met situations where the package was bragging about not using the Kr85, but in case it was used, there was no marking anywhere. So I would expect the lamp once had the Kr85 aid, but it decayed so it is missing now.
Does exposure to direct sunlight (real direct, the UV is what matters) make any difference?

Wiring is within half a metre from ballast to lamp, another 70W CMH lamp works perfectly fine and starts immediately on both ballasts.

Insulation glass on the base is intact with no signs of flashovers on either the socket or screw base, also I'm using a 4 KV rated HID socket. I have yet to try putting the lamp in direct sunlight, but I did try a xenon camera flash (external one for DSLR's), several times on full power to no avail. 

The thing with the external light is, it has to be present EXACTLY at the moment the ignitor is generating its pulse. And few times consecutively. I would see that as pretty difficult to achieve with the flash, but the sun shines constantly...

My very first QMH pulse start lamp, a 70w GE one, nearly always took from a few seconds to over a minute to start, with the ignitor buzzing away the entire time. Otherwise it worked normally. No other lamp I had was this bad. But that one almost always had the delayed start.

I think it's not uncommon as I've seen lamps in commercial settings that do the same. I also have a self-ballasted CMH lamp that sometimes takes a while to strike. It's not really a big deal for dusk-dawn outdoor lighting but can be a bit annoying for indoor lighting or something that is switched more frequently than once a day. I'm sure it wears out the ignitor faster too.