By the way, maybe these log messages relate to the fact that the PowerPC workers are just getting stuck (instead of being able to invoke kdump and reboot like the x86_64 worker I've run the same stress test on): ``` [535706993019,3] OPAL: Trying a CPU re-init with flags: 0x2 [536219455903,3] OPAL: CPU 0x409 not in OPAL ! ``` It looks like the kernel tries to do something with the CPU that needs to be done differently depending on whether the machine is running in OPAL mode or not and the kernel does it the wrong way. This is likely just bad post-mortem handling, though. The real problem is likely something different and likely not even related to the stress test I've been running. It nevertheless shows that Linux's PowerPC support doesn't seem very stable in general.