Comment 5 for bug 1652185

Google sent me here. I know I'm not ubuntu. Sorry. I experienced this issue on these:

1. debian backports linux-image-4.9.0-0.bpo.1-amd64 4.9.2-2~bpo8+1
2. linux-image-4.6.2 (custom compiled)

I got around these issues by booting this debian kernel : linux-image-3.16.0-4-amd64 3.16.39-1+deb8u1

And then I was able to use the system long enough to discover the drive was bad. Once replacing the failed drive (which appears to be what exploited this bug), I booted back into linux-image-4.9.0-0.bpo.1-amd64 4.9.2-2~bpo8+1

Once in 4.9 I ran the appropriate mdadm commands to get the array to start rebuilding. The array is currently rebuilding in 4.9 and the issue appears to gone. (before the server wouldn't stay up for 3 minutes. uptime is now 22min).