| *** sshnaidm is now known as sshnaidm|afk | 04:39 | |
| cenne | good morning janders o/ | 05:13 |
|---|---|---|
| * TheJulia needs more coffeeee | 05:38 | |
| cenne | wow.. you're still up. | 05:45 |
| cenne | All the best TheJulia, for what you are doing. | 05:53 |
| TheJulia | I'm chairing the Open Infra Foundation board meeting right now | 05:54 |
| cenne | oh cool! ++ | 06:04 |
| TheJulia | and soon I'll be going to bed | 06:05 |
| * TheJulia is normally in bed by 9pm | 06:05 | |
| TheJulia | it is 11 | 06:05 |
| iurygregory | good morning Ironic =) | 06:33 |
| janders | hey cenne TheJulia iurygregory o/ | 06:33 |
| iurygregory | hey janders o/ | 06:34 |
| TheJulia | Nini | 06:34 |
| TheJulia | Corgi and feline overlords are inches away saying time to sleep. | 06:34 |
| arne_wiebalck | Good morning iurygregory janders cenne and Ironic! | 06:34 |
| arne_wiebalck | Oh, and TheJulia ! | 06:34 |
| * iurygregory agrees with corgi and feline overlords | 06:35 | |
| iurygregory | hey arne_wiebalck o/ | 06:35 |
| janders | TheJulia I think Corgi and Feline Overlords are right! | 06:36 |
| TheJulia | Zzzzzzzzzzzz | 06:38 |
| cenne | good night TheJulia. | 06:39 |
| cenne | hey iurygregory arne_wiebalck janders o/ | 06:39 |
| iurygregory | hey cenne o/ | 06:39 |
| opendevreview | Will Szumski proposed openstack/bifrost master: Allow HTTP headers to be customised for IPA image download https://review.opendev.org/c/openstack/bifrost/+/798825 | 10:28 |
| dtantsur | good morning folks | 10:44 |
| iurygregory | good afternoon dtantsur =) | 10:44 |
| *** sshnaidm|afk is now known as sshnaidm | 10:53 | |
| janders | hey dtantsur o/ | 11:51 |
| mnaser | TheJulia: when you have as second, what is the intended replacement of `baremetal_admin` in the new rbac model? in this case, a system scope admin => admin of the whole cloud, but what if i wanted an ironic admin only? | 12:05 |
| janders | see you tomorrow Ironic o/ | 12:41 |
| opendevreview | Derek Higgins proposed openstack/ironic master: Skip port create if MAC is blank https://review.opendev.org/c/openstack/ironic/+/798844 | 12:50 |
| ajya | TheJulia, rpittau|afk & others, posted a story for sushy for discussion https://storyboard.openstack.org/#!/story/2009017 | 12:51 |
| dtantsur | ajya: I don't have objections to that | 12:55 |
| ajya | dtantsur: thanks | 12:57 |
| opendevreview | Derek Higgins proposed openstack/ironic master: Skip port create if MAC is blank https://review.opendev.org/c/openstack/ironic/+/798844 | 13:29 |
| TheJulia | ajya: no objections as long as we don't go walking the api entire api to figure out what we can have present... or not on the object. | 14:10 |
| dtantsur | morning TheJulia | 14:21 |
| TheJulia | good morning dtantsur | 14:26 |
| TheJulia | I would appreciate a review on https://review.opendev.org/c/openstack/ironic-python-agent/+/798732 It is train only, to fix my screw up | 14:27 |
| ajya | TheJulia: thanks | 14:29 |
| JayF | TheJulia: is that OK to merge into a stable branch? | 14:29 |
| TheJulia | JayF: I think so, its never going to be released again so it is patches to fix vendors downstream branches at this point | 14:30 |
| JayF | oh, it's EM. Hmm that does change the calculus some. | 14:30 |
| JayF | I really don't like that patch, but I can pretend I never saw it if you want :) | 14:30 |
| TheJulia | the change that introduced it was also an em patch | 14:30 |
| JayF | yeah IMO I would've reverted the backported patch with inconsistent behavior, and probably declared it unbackportable | 14:31 |
| JayF | which is not a good solution either | 14:31 |
| JayF | but seeing if python2; bail in code that we advertised as supporting python 2 seems pretty rough too | 14:31 |
| TheJulia | its all kind of awful since it didn't break in gate, but python internal codepoint checking blew up when it hit the utf-16 encoding marker on rhel in py36 because it needs explicit encoding satement | 14:31 |
| JayF | Oh, I understand the failure semantics, and it's no fun :-( | 14:31 |
| JayF | TheJulia: internet says `io.open` exists in py27 standard library, and gives you an encoding param | 14:32 |
| TheJulia | My focus, atm is to just fix broken downstream stable/train, so I could try to pen the file separate ways if we want. Just seems like supporting py2 still in any way is a pita | 14:32 |
| JayF | TheJulia: apparently it's slow and python-only in py2 but fast and uses the C path in py3 | 14:32 |
| TheJulia | JayF: orignal attempt to pass utf-16 encoding blew up | 14:33 |
| TheJulia | wait, io.open allows for encoding=utf-16? | 14:33 |
| JayF | TheJulia: my concern is actually amplified by this being IPA code... that warning may not be seen by folks | 14:33 |
| JayF | TheJulia: that's what that stackoverflow said | 14:33 |
| JayF | https://stackoverflow.com/questions/10971033/backporting-python-3-openencoding-utf-8-to-python-2 | 14:34 |
| TheJulia | does it honor the utf-16 delimieter marker? | 14:34 |
| * TheJulia reads | 14:34 | |
| JayF | all I theoretically know is from that stack overflow answer | 14:34 |
| JayF | which means I don't actually know it until I read some docs :) | 14:34 |
| TheJulia | hmm, that should be easy | 14:35 |
| JayF | the answer by YaOZl looks to be straight copy+pastable | 14:35 |
| TheJulia | well, version dependent though | 14:35 |
| JayF | true but I don't mind if py2; blah | 14:35 |
| JayF | I only mind just punting in that case :) | 14:35 |
| JayF | if six.PY2 is not that abnormal to see in 2/3 supporting code | 14:36 |
| TheJulia | no, but I think we started to strip six out around this time. RDO really frustrates me by sypporting py27 | 14:36 |
| JayF | I mean, it is what it is. If we said we were gonna support a thing, regardless of if we personally know of any customers, we should support it | 14:37 |
| JayF | and especially not drop that support when least expected; in EM | 14:37 |
| TheJulia | well, its complex | 14:42 |
| TheJulia | I'll give it a shot. I think since my problem is only py3 native, it shouldn't e a big deal really. | 14:42 |
| dtantsur | you can always experiment with your BOOTX64.CSV locally | 14:43 |
| TheJulia | That requires me to have py2 installed | 14:44 |
| TheJulia | :( | 14:44 |
| TheJulia | I surely have a VM installed someplace | 14:44 |
| dtantsur | it's still available on Fedora | 14:44 |
| dtantsur | if you give me a code snippet, I'll run it | 14:44 |
| TheJulia | k | 14:45 |
| * TheJulia still feels asleep | 14:45 | |
| dtantsur | TheJulia: http://paste.openstack.org/show/807076/ | 14:48 |
| TheJulia | dtantsur: nice! *looks* like it did the needful. Can you have it spit it out by character number real quick? :) | 14:49 |
| JayF | \o/ | 14:50 |
| dtantsur | like this? http://paste.openstack.org/show/807077/ | 14:51 |
| JayF | Next time, just paste the broken code into the github AI and teach it how to replace you | 14:51 |
| dtantsur | :D | 14:51 |
| JayF | BTW; that thing seems super problematic | 14:51 |
| TheJulia | dtantsur: that works just as well | 14:51 |
| JayF | I am not touching it with a 50000 foot pole | 14:51 |
| JayF | and I hope other open source contributors take the same approach | 14:51 |
| dtantsur | yeah, it's very questionable | 14:51 |
| TheJulia | JayF: yeah, heard about that and it seems... kind of disturbing | 14:52 |
| TheJulia | what if the AI wants to code in a bunch of backdoors and you just trust it | 14:52 |
| JayF | TheJulia: I'm not even going there; I'm thinking copyright/licensing issues | 14:52 |
| TheJulia | oh jeeze yeah | 14:52 |
| TheJulia | didn't even think of that | 14:52 |
| JayF | TheJulia: for example; if it pulls one of our cool hardware tricks out of IPA while someone is using that IDE for a commercial competetor to IPA, is that OK? | 14:52 |
| JayF | yeah, that's why I mention it here; not just for the chat but to say a little bit of !!! | 14:53 |
| JayF | and discourage people from muddying the waters in the upstream project | 14:53 |
| JayF | I'm less worried about AI backdooring, or things like that, because just like with stack overflow, if you keep the code it gives you without personally validating it you're basically committing engineering malpractice anyway | 14:53 |
| * TheJulia watches py27 tests run again | 14:58 | |
| TheJulia | wheeee they passed | 14:58 |
| cenne | :) | 15:04 |
| cenne | tests going green is a good feeling | 15:04 |
| JayF | until you find a bug AFTER the tests turn green | 15:06 |
| JayF | and you lose confidence in them | 15:06 |
| JayF | /o\ | 15:06 |
| cenne | hehe. that's happened to me just now. | 15:07 |
| JayF | That's when you find how to run the test so it fails | 15:07 |
| JayF | gotta answer the question why were tests OK when the code wasn't | 15:07 |
| cenne | Wrote new tests that dtantsur suggested. felt nice even seing red. atleast the bug got found. | 15:07 |
| cenne | yup. lesson being learned. | 15:08 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent stable/train: Train only - Fix py3 support for bootloader default config load https://review.opendev.org/c/openstack/ironic-python-agent/+/798732 | 15:11 |
| JayF | in many situations; especially in IPA early days, I'd write the unit tests all first because a full round trip on real hardware took so long | 15:11 |
| JayF | so just "mock" the executes to return whatever that system command would run, and exercise the code through unit tests | 15:12 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent stable/train: Train only - Fix py3 support for bootloader default config load https://review.opendev.org/c/openstack/ironic-python-agent/+/798732 | 15:12 |
| JayF | that's a lot harder for API code I think | 15:12 |
| TheJulia | It is | 15:13 |
| TheJulia | *but* if you do it, once you get things at least wired through it makes things go a lot faster | 15:13 |
| dtantsur | writing a unit test can be a good way to confirm/reject a hypothesis | 15:14 |
| TheJulia | I think for the loading default bootlaoder stuff originally, I actually wrote the test with bfournie on the call first before writing a line of code | 15:14 |
| TheJulia | well, a line of code in image.py | 15:14 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP Scoped RBAC Devstack Plugin support https://review.opendev.org/c/openstack/ironic/+/778957 | 15:54 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent stable/train: Train only - Fix py3 support for bootloader default config load https://review.opendev.org/c/openstack/ironic-python-agent/+/798732 | 15:57 |
| TheJulia | JayF: pep8 fix | 15:57 |
| TheJulia | ^^ | 15:57 |
| JayF | I had it +2'd before you got the ^^ out :P | 15:58 |
| TheJulia | nice | 16:03 |
| * TheJulia suspects her devstack changes are all good now for scope enforcement, at least for cross-service between ironic/ironic-inspector | 16:12 | |
| arne_wiebalck | bye everyone, see you tomorrow o/ | 16:20 |
| TheJulia | goodnight arne_wiebalck | 16:24 |
| opendevreview | cenne proposed openstack/ironic master: Add `boot_mode` and `secure_boot` to node object and expose in api https://review.opendev.org/c/openstack/ironic/+/797055 | 16:34 |
| cenne | gn arne_wiebalck | 16:34 |
| dtantsur | I'm also leaving for a day, see you tomorrow | 16:47 |
| cenne | gn dtantsur | 16:57 |
| opendevreview | Julia Kreger proposed openstack/ironic-inspector master: Add rbac scope enforcement handling to devstack plugin https://review.opendev.org/c/openstack/ironic-inspector/+/798359 | 17:41 |
| TheJulia | stevebaker: question of the day, do we need to do anything with metalsmith w/r/t secure rbac work | 18:20 |
| TheJulia | stevebaker: https://089e4f224791bf0b3f34-27a42f979dc9642e645167ee17c588d5.ssl.cf1.rackcdn.com/778957/27/check/ironic-inspector-tempest-rbac-scope-enforced/df7fce0/testr_results.html <-- inspector + ironic talking using system scoped access works \o/ | 19:00 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Scoped RBAC Devstack Plugin support https://review.opendev.org/c/openstack/ironic/+/778957 | 19:25 |
| TheJulia | just commit message updates | 19:26 |
| stevebaker | morning | 20:10 |
| stevebaker | TheJulia: metalsmith as used by tripleo, I don't think so. All actions are done with one admin user | 20:11 |
| TheJulia | I guess they would just shift that to a system-admin user then | 20:11 |
| TheJulia | worst comes to worst, I intend to force most jobs over to enforced scope at some point | 20:12 |
| TheJulia | stevebaker: fyi, I swapped out the train fix to take a different path, just needs another review | 20:14 |
| stevebaker | I will check that out | 20:31 |
| JayF | TheJulia: I'm ready to +2 https://review.opendev.org/c/openstack/ironic/+/792274 -- do you want it +A'd too, or should it wait / merge many of those related changes as a block | 20:55 |
| JayF | btw that looks 5000% more readable after the sqlalchemy maintainer-requested changes | 20:55 |
| TheJulia | JayF: Yeah, once I figured out how to do it and keep us from doing more work :) | 20:56 |
| JayF | Also, do we intend on backporting the DB perf changes? | 20:56 |
| TheJulia | JayF: I intend to backport them, so I think you can merge at will, none of the stack is breaking behavior wise | 20:56 |
| JayF | +1 awesome, will save me the effort if we can get them backported upstream :D | 20:57 |
| TheJulia | JayF: I can likely backport them to train, but I'm not sure I'll pick them up downstream on train fwiw | 20:58 |
| JayF | TheJulia: the code now looks a lot like the more optimized sqlalchemy code we churned out at $lastJob | 20:58 |
| TheJulia | maybe merging as a block will make backporting easier | 20:58 |
| TheJulia | at least tracking wise | 20:59 |
| JayF | it's up to you, I also know dmitry had a strong -1 on it, and IDK if the changes resolved his concerns (looks like it to me?) | 21:01 |
| JayF | so I'll just drop a +2 on it, and we can block-merge, I think that's the better idea anyway | 21:01 |
| TheJulia | at a minimum, I need to get them backported to wallaby because the rbac stuff added more overhead and possibilities to make things more painful performance wise accidently | 21:02 |
| TheJulia | wfm | 21:02 |
| TheJulia | I did, afaik, take care of everything Dmitry was objecting to | 21:02 |
| JayF | we should get it back as far as possible if we have strong tests to ensure no regression | 21:02 |
| TheJulia | I just think he has been too busy to get back to reviewing it | 21:02 |
| JayF | that'd be a really nice bonus for someone tracking stable lol | 21:02 |
| TheJulia | ++ it is all api field contract based so I think the testing I add in along the way does the sufficent testing required and all of the tempest tests pass without blinking and nova just works which I consider the ultimate test | 21:03 |
| TheJulia | oh yeah, I feel personally responsible for making the performance worse, so... I now feel the need to make it better :) | 21:04 |
| TheJulia | silly head | 21:04 |
| JayF | You gotta flip that | 21:04 |
| JayF | since everyone made it slow over the course of like 8 years | 21:05 |
| JayF | it's not your fault :P | 21:05 |
| JayF | what really hurts is when I fix something in IPA that's obviously been there since it was the teeth-agent | 21:05 |
| JayF | lol | 21:05 |
| JayF | no buck passing there | 21:05 |
| TheJulia | We've been focused on features as a community so much becasue that was the immediate needful that we never really stoped to quantify things at the end | 21:05 |
| TheJulia | stopped | 21:05 |
| * TheJulia really likes the "lets not do a community goal and just focus on improving pain points" thread | 21:06 | |
| JayF | Bluntly, openstack is so CI driven that it gets very easy to ignore scale issues unless you have a loud open operator (like arne_wiebalck) pointing at the issues | 21:06 |
| TheJulia | ++++++++++ | 21:06 |
| TheJulia | It is, and CI systems can have variable performance | 21:06 |
| TheJulia | at least, cloud based CI | 21:06 |
| TheJulia | iurygregory: tomorrow, I could use like 20 minutes of your assistance looking at puppet-ironic | 21:26 |
| TheJulia | well, your today once you finally read this my tomorrow | 21:26 |
| opendevreview | Merged openstack/ironic master: Refactor deploy_utils.validate_image_properties https://review.opendev.org/c/openstack/ironic/+/797875 | 22:37 |
| janders | good morning Ironic o/ | 23:05 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!