Oak tests update patches, by triggering nightlies through self-serve. That means that it only triggers nightlies which built during the last two weeks. As a result, it knows to trigger the "Android Armv6 oak nightly" and "Android oak nightly" and "Android X86 oak nightly" which do not exist, but doesn't know to trigger the three of those with "2.2" inserted after the word "Android". (Extra credit for caring about whether a new name for static analysis nightlies and whatever replaced b2g_oak_ics_armv7a_gecko nightly exist, and sendchanging them too - I don't care, I just kill those two whenever I notice them pending. And of course, super bonus extra credit for deleting every single one of those two and all three Android ones for the past two weeks, which would let me stop killing all those pending jobs that will never run, forever.)

Looks like this has been done already. For 'Android 2.2 oak nightly' a job was forced in buildbot at Friday 21:47 PT, and then more nightlies were scheduled automatically at 5am Saturday. That's on the tip d7bd7ce48780 which landed Friday afternoon. Same story for 'Android 2.2 Armv6 oak nightly' and 'Android 4.2 x86 oak nightly'. Forcing may not have been necessary if waiting for 5am Sat was fine. There's no static analysis nightly, just dep builds. b2g_oak_win32_gecko nightly and the localiser build happened as above, only 1430 ish via self-serve by bbondy so that's unchanged & working.