| *** dhill_ has quit IRC | 00:29 | |
| *** yolanda has joined #openstack-dib | 08:44 | |
| *** tflink has quit IRC | 09:01 | |
| *** tflink has joined #openstack-dib | 09:02 | |
| *** yolanda_ has joined #openstack-dib | 09:03 | |
| *** dhill_ has joined #openstack-dib | 14:26 | |
| ianw | tonyb: any thoughts on https://review.openstack.org/#/c/536852 and ppc64le v el ? i feel like we've discussed this before | 21:54 |
|---|---|---|
| openstackgerrit | Merged openstack/diskimage-builder master: Remove architecture rules on lshw dependency in ironic-agent https://review.openstack.org/526212 | 22:07 |
| tonyb | ianw: I'll comment on the review, it's basically correct if we're comparing to the distro arch (which I think we're doing). At some stage it'd be cool split ditroarch from kernel arch and provide functions to check both. That'd handle amd64 vs x86_64 and ppc64le vs ppc64el | 22:08 |
| tonyb | ianw: selfishly I'd like to see it rebased on https://review.openstack.org/#/c/526212/ | 22:09 |
| tonyb | Oh which you've just approved! | 22:09 |
| tonyb | ianw: ANy chance you can revisit https://review.openstack.org/#/c/532279/4 ? | 22:28 |
| ianw | tonyb: yeah, i think we can still do better on 532279 | 22:30 |
| tonyb | ianw: okay | 22:30 |
| ianw | let me see more closely ... | 22:31 |
| tonyb | ianw: If you have time that'd be great. | 22:31 |
| ianw | my strong preference is to keep the hack isolated to one place | 22:32 |
| tonyb | ianw: okay I get that but I kinda thought that's what PS4 did. | 22:33 |
| ianw | yes, but i don't think merge_into_config is the right place ... that's a sort of hack function due to us supporting old environment variables from the prehistoric era | 22:34 |
| tonyb | ianw: Ahh okay | 22:34 |
| * ianw context switches back in how all this works :) | 22:34 | |
| tonyb | ianw: sorry (but thanks) | 22:35 |
| *** pmannidi has joined #openstack-dib | 22:41 | |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Export DIB_ROOT_LABEL from final state https://review.openstack.org/532279 | 23:35 |
| ianw | tonyb: ^ ok, that's what i'm thinking ... are you able to try that out in the exact problem case? | 23:35 |
| tonyb | ianw: I am. takes me about 30mins to reproduce it. | 23:37 |
| tonyb | ianw: I'll get back to you abotu then ;P | 23:37 |
| tonyb | ianw: Actually it''l be about twice that so I can do A/B testing. | 23:41 |
| ianw | that's ok, i'm not going anywhere :) | 23:45 |
| ianw | between ppc64le/el, amd_64/x86_64, aarch64/arm64 have we collectively managed have history repeat time after time? | 23:47 |
| tonyb | ianw: Yup, the ppc64le/el is for different reasons to amd_64/x86_64. | 23:50 |
| tonyb | ianw: I don't really know the reasons/history for aarch64/arm64 | 23:51 |
| tonyb | ianw: the "right" solution is to decpuple the kernel arch from the distro arch and then use them appropriately | 23:51 |
| tonyb | Am I running on a 64-bit ppc box in LE mode vs 'what does my distro call the CPU I'm running on' | 23:52 |
| tonyb | I'm a bit on the fence as to whether we ever really need that second check but I could be convinced either way. | 23:53 |
| tonyb | ianw: If I get enough of OpenStack running on ppc64le I | 23:53 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Don't install dmidecode on Fedora ppc64le https://review.openstack.org/536852 | 23:54 |
| tonyb | 'll take a stab at cleaning it all up but that's a big patch series and a cultural chnage for the core team | 23:54 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!