The two MH lamp failures I experienced were-a 250WMH lamp in an enclosed fixture-the lamp made a muffled "boom" and none of the pieces escaped out of the bulb.Another was a 100W CMH lamp and a piece came out of the bulb but didn't leave the fixture-was enclosed.Again a muffled "pop" sound when the bulb went.

It mostly ends up this way, because the arctube tend to ultimately fail at tge moment of the greatest stress in the material and that uses to be during initial phases of the warmup, when there is not yet the high pressure (to pressurize, it needs the coldest spot to arm up, so when it finally does, it aleviates the stresses, so make tge rupture a bit less likely).
but there are not that rare cases, when the arctube blows when hot.
One most prominent cause is a ballast failure, heavily overloading the lamp as a result. When this happens to a relatively new kamp with still very strong arctube, it allows the pressure to build up really high and then when the arctube gives up, the explosion is really violent.
Other common situations when HIDs tend to blow up is, when it is operated 24/7, so the pressure is there all the time without having to transition through the stress stages, so when the arctube weakens further, it blows out this full pressure.
And there are manufacturing defects, where the arctube blows without warning even at full pressure, if the defect does not affect the spot exposed to the warmup stress. The problem is, unlike the defects affecting the stressed regions, these wont get activated at the end of line test (the ones weakening the stressed parts cause it to blow up during the runup test, so affected bulb blows at the manufacturers test head and never get to you), so are very likely to pass.