This revision has caused a strange issue with both testbots. At kernel32:console crash, when testbot restarts vm, the test suite seems to begin right from the start, instead of next kernel32 test group. It happens subsequently whenever testbot reaches console test and crashes, leading to an ugly loop.