I've been able to successfully execute the following upstream tests which mix non-swift usage (standalone), swiftless glance usage, and usage with swift in integrated configurations. All of these were tested *with* and without the conductor_always_verifies_image option set.
The major variation in these jobs are largely a variety of scenarios and interfaces, so overall I have good confidence that the ironic patch, as-is, is not going to have any major issues when dropped into the gate (when we get to that point).
The team can discuss further test jobs, but we should repeat tests of the "direct" deployment interface (ironic-standalone-redfish and the tempest jobs will exercise this) when we have a IPA patch in a ready state.
Furthermore, I took Jay's script, modified slightly and repeated tests, and did discover an issue as related to disqualification of images based upon user supplied input of the image_disk_format compared to what we have received url content wise. I'll upload a new revision shortly, but the tl;dr is all of the expected failures failed, and all of the good images succeeded. In the script I'll also upload, I denote what is expected. I also tested against, from a good measure standpoint of additional images just to ensure we have our bases covered ultimately ensure the fix is in the best shape possible.
I've been able to successfully execute the following upstream tests which mix non-swift usage (standalone), swiftless glance usage, and usage with swift in integrated configurations. All of these were tested *with* and without the conductor_ always_ verifies_ image option set.
ironic- standalone- redfish standalone- anaconda tempest- bios-redfish- pxe tempest- ipa-wholeisk- bios-ipmi- direct- dib
ironic-
ironic-
ironic-
The major variation in these jobs are largely a variety of scenarios and interfaces, so overall I have good confidence that the ironic patch, as-is, is not going to have any major issues when dropped into the gate (when we get to that point).
The team can discuss further test jobs, but we should repeat tests of the "direct" deployment interface (ironic- standalone- redfish and the tempest jobs will exercise this) when we have a IPA patch in a ready state.
Furthermore, I took Jay's script, modified slightly and repeated tests, and did discover an issue as related to disqualification of images based upon user supplied input of the image_disk_format compared to what we have received url content wise. I'll upload a new revision shortly, but the tl;dr is all of the expected failures failed, and all of the good images succeeded. In the script I'll also upload, I denote what is expected. I also tested against, from a good measure standpoint of additional images just to ensure we have our bases covered ultimately ensure the fix is in the best shape possible.