Comment 44 for bug 1640547

@Colin, one final question... I ran an updated stress-ng (0.07.16-1ppa1) on a Xeon Phi system with a TON of cores that was able to reproduce these bugs that we found on power. Note, that version is where we copied stress-ng 0.07.16 to the cert PPA and built it for Xenial since that version doesn't currently exist in Xenial but is needed to resolve these issues for Xenial certs.

Anyway, using a combination of that and your modifications to disk_stress_ng, I was able to run the tests on the Phi system with no failures except for one that LOOKS like a failure, but appears to succeed anyway:

http://pastebin.ubuntu.com/23996480/

The disk in question is:
Disk /dev/sdb: 111.8 GiB, 120034123776 bytes, 234441648 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: dos
Disk identifier: 0xb732291a

Device Boot Start End Sectors Size Id Type
/dev/sdb1 2048 234440703 234438656 111.8G 83 Linux

It's formatted w/ ext3.

This is sync-file run manually with -v enabled:

sudo stress-ng --aggressive --verify --timeout 240 --temp-path /mnt --sync-file 0 --hdd-opts dsync --readahead-bytes 16M -k -v --log-file stress-test

http://paste.ubuntu.com/23997177/

So my concern is whether all those error messages indicate a failure that isn't being treated as such by stress-ng, or if they're benign and expected (and if so, they're confusing because it looks like something bad happened to me).