| opendevreview | Iury Gregory Melo Ferreira proposed openstack/ironic master: RedfishFirmwareInterface - Unit Tests & More logs https://review.opendev.org/c/openstack/ironic/+/903379 | 02:11 |
|---|---|---|
| opendevreview | likui proposed openstack/ironic-lib master: Remove compatibility with Python < 3.8 https://review.opendev.org/c/openstack/ironic-lib/+/905811 | 02:36 |
| opendevreview | likui proposed openstack/ironic-prometheus-exporter master: Update python classifier in setup.cfg https://review.opendev.org/c/openstack/ironic-prometheus-exporter/+/905812 | 02:38 |
| opendevreview | likui proposed openstack/python-ironicclient master: Update python classifier in setup.cfg https://review.opendev.org/c/openstack/python-ironicclient/+/905813 | 02:40 |
| opendevreview | likui proposed openstack/ironic-ui master: Update python classifier in setup.cfg https://review.opendev.org/c/openstack/ironic-ui/+/905815 | 02:44 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/2023.2: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905725 | 04:30 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/2023.1: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905726 | 04:31 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/zed: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905832 | 08:10 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/yoga: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905833 | 08:11 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/xena: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905834 | 08:11 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-python-agent-builder stable/wallaby: Drop TripleO job https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/905835 | 08:13 |
| rpittau | good morning ironic! o/ | 09:07 |
| masghar | Morning! | 09:11 |
| opendevreview | Dmitry Tantsur proposed openstack/bifrost master: Switch to snponly and stop copying the unused ipxe.pxe https://review.opendev.org/c/openstack/bifrost/+/905844 | 10:00 |
| dtantsur | TheJulia: we somehow missed bifrost ^^ | 10:00 |
| iurygregory | good morning Ironic | 11:01 |
| opendevreview | Merged openstack/ironic-lib master: Raise on non-existing file in qemu_img_info https://review.opendev.org/c/openstack/ironic-lib/+/905612 | 11:11 |
| *** ravlew is now known as Guest14423 | 11:33 | |
| *** ravlew2 is now known as ravlew | 11:37 | |
| opendevreview | Merged openstack/ironic master: Revert "Revert "RBAC: Fix allocation check"" to use Unauthorized https://review.opendev.org/c/openstack/ironic/+/905784 | 13:14 |
| Sandzwerg[m] | Morning Ironic o/ | 13:57 |
| Sandzwerg[m] | A Question to virtual media together with glance & ISO: Do I need to configure anything else apart from setting deploy_iso=<glance image id> in the driver-info of the node? Because I now tried it with a newer dell node instead of the old cisco and it still seems to skip the virtual-media part. But I have the feeling I'm missing something or do something wrong. Our ironic is still on xena in case that should matter | 14:00 |
| TheJulia | Good morning | 14:14 |
| * TheJulia starts sipping coffee | 14:14 | |
| TheJulia | Sandzwerg[m]: so deploy_iso is much more "this is a pre-existing iso image" to use, where as we prefer to start with a kernel/ramdisk file and master an ISO on the fly as we include configuration in the ISO. If I'm recalling correctly, deploy_iso is much more a hold over from earlier virtual media drivers | 14:17 |
| Sandzwerg[m] | Hope you're healthy again | 14:17 |
| TheJulia | I mean, it should work, but it might not be the preferred path/behavior | 14:17 |
| TheJulia | work up this morning feeling human! So, so far so good, albeit a little cold for me in the desert this morning | 14:18 |
| Sandzwerg[m] | hmmmkay. Maybe I need to look into the build-iso on the fly then. I just build our IPA as ISO instead of bothering with installing the requirements to build ISOs on the fly. But so far I have the impression that it does not try to boot from the ISO (or any other kind of virtual media) instead it just does the default boot order which does not include virtual media | 14:20 |
| Sandzwerg[m] | Need to check in the bios if I can add virtual media to the boot order 🤔 | 14:20 |
| Sandzwerg[m] | And feeling human sounds like a good start | 14:21 |
| TheJulia | Sandzwerg[m]: Are you in bios mode or uefi mode? I guess the thing is we explicitly ask for the CD device to be booted from | 14:23 |
| TheJulia | or DVD device | 14:23 |
| Sandzwerg[m] | UEFI | 14:23 |
| TheJulia | I guess it would help to figure out where things are *actually* falling sideways at | 14:23 |
| TheJulia | in your logs, you should see us assering a BootSourceOverrideTarget | 14:24 |
| TheJulia | asserting | 14:24 |
| Sandzwerg[m] | yeah but that's still a mistery to me. Should something be in the conductor log? | 14:24 |
| Sandzwerg[m] | Allright let's dig goes into the log mines https://www.youtube.com/watch?v=ytWz0qVvBZ0 | 14:24 |
| TheJulia | Sandzwerg[m]: yeah, that will be in the conductor log | 14:26 |
| TheJulia | look for errors from "sushy" | 14:26 |
| TheJulia | +2+A to the video | 14:26 |
| TheJulia | My wife has played that video far too many times in the home office | 14:28 |
| Sandzwerg[m] | There is nothing like "to manz times" when it comes to that video 👼 | 14:28 |
| Sandzwerg[m] | OK the fact that did not set the node to virtual-media when I was 100% sure I did is a good explanation why it didn't work so far... | 14:31 |
| Sandzwerg[m] | hmm ok that is Interesting. I set it to virtual media, boot, and it's back at ipxe... | 14:32 |
| Sandzwerg[m] | I assume that should not happen? | 14:32 |
| TheJulia | no, that should not really happen | 14:33 |
| TheJulia | on the node boot_interface field? | 14:33 |
| Sandzwerg[m] | yeah | 14:33 |
| TheJulia | AFAIK there is no reason why that would have just "changed" | 14:34 |
| TheJulia | at least, in ironic itself | 14:34 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent stable/2023.2: Add missing headers to the inspection callback https://review.opendev.org/c/openstack/ironic-python-agent/+/905739 | 14:34 |
| opendevreview | Dmitry Tantsur proposed openstack/sushy stable/2023.2: Handle a different related properties for missing TransferProtocolType https://review.opendev.org/c/openstack/sushy/+/905740 | 14:35 |
| Sandzwerg[m] | yeah seems very strange. BTW regarding boot mode: the node itself is set to UEFI but we haven't set the boot mode in ironic | 14:35 |
| Sandzwerg[m] | I need to look into this tomorrow, need to run some errands and probably have no more time to today. Thanks for the help so far \o/ | 14:35 |
| TheJulia | Sandzwerg[m]: as long as ironic's understanding matches, it should be fine then | 14:37 |
| TheJulia | it will try to assert whatever it thinks the state/mode is as well | 14:37 |
| TheJulia | so at the end of the day it just noops | 14:37 |
| TheJulia | or at least *should* noop | 14:37 |
| Sandzwerg[m] | For us we just want to boot via uefi instead of bios but don't need to set any specific uefi stuff. Just write the whole disk image to disk and we're fine | 14:57 |
| Sandzwerg[m] | And I think setting uefi as boot mode would go in that direction but I would need to look it up | 14:58 |
| TheJulia | o/ can I get another review for https://review.opendev.org/c/openstack/ironic-inspector/+/905353 please? | 15:00 |
| opendevreview | Hervé Beraud proposed openstack/ironic master: bump eventlet to latest version that support python 3.12 https://review.opendev.org/c/openstack/ironic/+/905887 | 15:00 |
| opendevreview | Hervé Beraud proposed openstack/ironic-inspector master: bump eventlet to latest version that support python 3.12 https://review.opendev.org/c/openstack/ironic-inspector/+/905888 | 15:01 |
| opendevreview | Hervé Beraud proposed openstack/ironic-python-agent master: bump eventlet to latest version that support python 3.12 https://review.opendev.org/c/openstack/ironic-python-agent/+/905889 | 15:02 |
| iurygregory | TheJulia, looking now | 15:03 |
| opendevreview | Julia Kreger proposed openstack/ironic stable/2023.2: Revert "Revert "RBAC: Fix allocation check"" to use Unauthorized https://review.opendev.org/c/openstack/ironic/+/905901 | 15:04 |
| opendevreview | Julia Kreger proposed openstack/ironic stable/2023.1: Revert "Revert "RBAC: Fix allocation check"" to use Unauthorized https://review.opendev.org/c/openstack/ironic/+/905902 | 15:05 |
| opendevreview | Hervé Beraud proposed openstack/networking-generic-switch master: bump eventlet to latest version that support python 3.12 https://review.opendev.org/c/openstack/networking-generic-switch/+/905921 | 15:10 |
| dtantsur | my first reaction to https://review.opendev.org/q/topic:%22bump-eventlet%22 is to -2 all of them.. is there any reason why my first reaction should not become my final reaction? | 16:07 |
| JayF | Why would that be your reaction at all? | 16:07 |
| JayF | I would -2 a reaction of that kind :D | 16:07 |
| dtantsur | JayF: that's not the appropriate way to bump version | 16:07 |
| dtantsur | do you suggest we change how requirements are managed? | 16:07 |
| JayF | are you in #openstack-oslo, I have been having this conversation in there with someone, too | 16:08 |
| JayF | we have it bumped in u-c already | 16:08 |
| dtantsur | sure, these two are not related | 16:08 |
| JayF | there is no other way to force a lower bound | 16:08 |
| dtantsur | fortunately, you don't need to force it? | 16:08 |
| JayF | except in each project individually aiui | 16:08 |
| JayF | You don't need to; I agree with the idea of proposing it everywhere due to the massive improvements/cleanups in eventlet | 16:08 |
| JayF | -1 with cause in a project would be a sensible thing. Rejecting out of hand is not sensible because of the significant increase in quality of the underlying library. We're saying "we don't want to support OpenStack on the unsupported, no CI tested versions of eventlet" which is a really good thing to say | 16:09 |
| dtantsur | THIS is the argument I would buy | 16:10 |
| JayF | This is 100% how I took it based on context that has been in the mailing list around eventlet | 16:10 |
| dtantsur | yeah, but the operators may not have this context | 16:12 |
| dtantsur | and eventlet updates are not to be taken lightly | 16:12 |
| JayF | Yeah, it was suggested in -oslo an email should go out about the updates | 16:13 |
| JayF | I was like "sure whatver, if you think so" | 16:13 |
| JayF | then you were like "WTF" in here literally 30 seconds later | 16:13 |
| JayF | the continuity between the two conversations was uncanny :) | 16:13 |
| TheJulia | heh | 16:13 |
| dtantsur | I'd rather have a release note, to be honest | 16:13 |
| dtantsur | maybe we can write one ourselves | 16:14 |
| TheJulia | I'm surprised to hear you say that | 16:14 |
| JayF | I am going to pair with adamcarthur5 for a bit this morning, after that if someone else hasn't, I'll email the list about the update. | 16:14 |
| TheJulia | since you've pushed back on other dependency updates being inremented being explicitly noted in renos. | 16:14 |
| JayF | I am torn on the idea of a release note; but I don't want hberard getting beat up for what's ultimately a good thing with bad PR :D | 16:14 |
| dtantsur | TheJulia: I have, but this for me is an exceptional case | 16:15 |
| TheJulia | fair | 16:15 |
| dtantsur | It's less of "We now require this", it's more of "Eventlet has been broken, and now we require the only fixed version" | 16:15 |
| dtantsur | My reasoning is usually "they'll see the update in the requirements". Here I want to convey the scope of the problem. | 16:15 |
| JayF | this will also say loudly to downstreams | 16:16 |
| dtantsur | Kind of a "fixes:" release note for Ironic changes | 16:16 |
| JayF | if you package openstack with old-eventlet | 16:16 |
| JayF | and you break things | 16:16 |
| * dtantsur hopes he makes sense | 16:16 | |
| JayF | here, have all the pieces | 16:16 |
| dtantsur | exactly | 16:17 |
| dtantsur | anyway, thanks for limiting my first reaction to the first reaction JayF :) | 16:18 |
| Sandzwerg[m] | <Sandzwerg[m]> "For us we just want to boot..." <- So setting virtual media also sets boot mode to uefi, then on deployment (or shortly before but that would be surprising, still needs more testing) the ironic service user sets the node to maintenance, sends a patch call und unsets maintenance | 16:31 |
| opendevreview | Verification of a change to openstack/ironic-inspector master failed: Fix RBAC access for service/admin user access https://review.opendev.org/c/openstack/ironic-inspector/+/905353 | 16:37 |
| opendevreview | Merged openstack/networking-generic-switch master: bump eventlet to latest version that support python 3.12 https://review.opendev.org/c/openstack/networking-generic-switch/+/905921 | 16:41 |
| opendevreview | Adam McArthur proposed openstack/ironic-lib master: [codespell] - Adding CI for codespell tox target https://review.opendev.org/c/openstack/ironic-lib/+/905624 | 16:44 |
| opendevreview | Adam McArthur proposed openstack/ironic-lib master: [codespell] - Adding CI for codespell tox target https://review.opendev.org/c/openstack/ironic-lib/+/905624 | 17:02 |
| rpittau | good night! o/ | 17:04 |
| opendevreview | Adam McArthur proposed openstack/ironic-python-agent master: [Codespell] Adding tox target to CI (Non-Voting) https://review.opendev.org/c/openstack/ironic-python-agent/+/905980 | 17:17 |
| JayF | Ah, we need to make sure eventlet .34.3 is in all supported openstack platforms before we can require it. That may not be likely :/ | 18:14 |
| TheJulia | Unlikely is more so the thing. | 18:40 |
| TheJulia | Sandzwerg[m]: nothing we would be doing sets a node to maintenance to patch the node, and unsets it, that really doesn't explain it. What is calling Ironic in that context? | 18:41 |
| JayF | Yeah, chatter in -tc makes this sound like it's more likely these changes won't land, but I get the impression either nobody read my ML post or does not want to engage in a longer form so the conversation is bifurcated instead of concentrated :( | 18:42 |
| TheJulia | The nature is going to be generally bifrucated because human can only hold so much context | 18:43 |
| JayF | yeah, but that's exactly why I was hoping we'd venue-swap to something friendlier to async, since the actual-changes were pushed by someone who I think (hope?) is asleep | 18:45 |
| TheJulia | hope++ | 18:47 |
| TheJulia | (but I think, yes, they should be afaik) | 18:47 |
| opendevreview | Julia Kreger proposed openstack/networking-generic-switch master: Revert "bump eventlet to latest version that support python 3.12" https://review.opendev.org/c/openstack/networking-generic-switch/+/905904 | 18:50 |
| TheJulia | I think there is also hope in terms of "we hope to move things on faster", and often not everyone is ready/able/capable to ride along at the same speed. | 18:50 |
| TheJulia | I'm all for pushing it as a req if "we as a project *know* people should just not be running the version of eventlet because it is inherently broken" | 18:51 |
| JayF | *I* feel that way. | 18:51 |
| JayF | It's not clear to me that is the project consensus. | 18:51 |
| JayF | And a good point was made in -tc that we are past C-2, it's maybe a bit late to make a change of this size. | 18:52 |
| JayF | and that argument hit me right in the ops | 18:52 |
| TheJulia | often humans will error on not changing until they are explicitly forced by some driving thing. Often it might be context | 18:53 |
| TheJulia | your raising very good points though, fwiw :) | 18:53 |
| TheJulia | "right in the ops" is a new one | 18:53 |
| TheJulia | rutro https://498c49910e2e429f6f6d-5ec4103c89b5d2feb0a0ca5aa5752932.ssl.cf1.rackcdn.com/905353/1/gate/openstack-tox-docs/320c0d4/tox/docs/2-commands%5B0%5D.log | 18:58 |
| JayF | hahahahahahahahthud | 19:00 |
| JayF | Can you put that into an eventlet issue? | 19:00 |
| TheJulia | sure, likely in a few minutes | 19:01 |
| TheJulia | I'm not sure I'm signed into gh on this laptop right now | 19:01 |
| TheJulia | nope, doesn't look like it, checking other browser | 19:01 |
| TheJulia | I don't have my password store handy eitehr at the moment (sitting in an urgent care lobby) | 19:02 |
| TheJulia | yeah, nope, it will have to wait until get back to the house | 19:02 |
| JayF | ack; we can race, I'm about to pair again with adamcarthur5 | 19:03 |
| JayF | look like IP_VERSION=4 doesn't block ipv6 use anymore | 19:03 |
| JayF | so I gotta find out what odes :D | 19:03 |
| TheJulia | nothing really does, I think devstack does 4+6 by default now | 19:04 |
| TheJulia | although did it break in an unexpected way? | 19:05 |
| TheJulia | also can't repro issue locally, apparently the new build wants to build yappi which won;t build due to a syntax issue according to gcc | 19:05 |
| JayF | He has a machine with no v6 access | 19:05 |
| JayF | and iP_VERSION=4 is set, it's still doing v6 stuff | 19:05 |
| TheJulia | that is bizzare | 19:06 |
| clarkb | devstack networking is independent of your v6 access. I don't find that bizarre | 19:08 |
| clarkb | you can have a fully working v6 stack between the hypervisor and the VMs that neutron connects | 19:08 |
| clarkb | and you can extend that to multinode testing via overlay networking. This is all important because the clouds we test on don't necessarily support v6 but we still need to test it | 19:09 |
| clarkb | basically whether or not you have a publicly routable v6 address is completely irrelevant to devstack | 19:09 |
| JayF | it's disabled by their VPS at a kernel level :/ | 19:09 |
| JayF | We figured it out, we're going to fix the doc and it'll be obvious | 19:10 |
| TheJulia | oh, becuase parameters still end up in the v6 style? | 19:12 |
| JayF | No; bitrotted doc | 19:12 |
| TheJulia | ahh | 19:12 |
| JayF | IP_ADDRESS vs IP_VERSION | 19:12 |
| JayF | inconsistently in our doc | 19:12 |
| JayF | IP_VERSION is correct | 19:12 |
| clarkb | disabling v6 at the kernel level is an interesting decision. Almost makes me wonder if the hosting provider doesn't provide adequate v6 isolation and they're faking it this way | 19:13 |
| clarkb | just hope no one figures out how to override the setting and start acting as a router :) | 19:14 |
| JayF | I do not know. Not my choice of provider and I'm trying to go with the flow of adamcarthur5's setup as best as possible :) | 19:14 |
| JayF | he has all sorta crazy devpoddy magic setup | 19:14 |
| JayF | oh, I think he could turn it on, but again I don't want to mess with his setup | 19:15 |
| clarkb | you need v6 to test a large numebr of things | 19:16 |
| clarkb | mostly neutron related | 19:16 |
| opendevreview | Adam McArthur proposed openstack/ironic master: Corrected IP_ADDRESS to IP_VERSION https://review.opendev.org/c/openstack/ironic/+/905996 | 19:16 |
| clarkb | again not publicly routable v6 just v6 on your sytstem | 19:16 |
| * TheJulia glares at the 67kB/sec download | 19:17 | |
| JayF | Yep, understood. For now just literally trying to get him to "hello world, this is an openstack API" so he gets to experience more of our world than setting up tox :D | 19:18 |
| TheJulia | err, ugh, my inability to repro is python 3.12 *and* the lack of packaging for other 3.11 assets compat assets and another 3.12 incompatibility | 19:30 |
| opendevreview | Merged openstack/networking-generic-switch master: Revert "bump eventlet to latest version that support python 3.12" https://review.opendev.org/c/openstack/networking-generic-switch/+/905904 | 19:55 |
| opendevreview | Merged openstack/ironic master: Corrected IP_ADDRESS to IP_VERSION https://review.opendev.org/c/openstack/ironic/+/905996 | 20:06 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Fix bug in devstack-guide: readd server create https://review.opendev.org/c/openstack/ironic/+/906001 | 20:06 |
| TheJulia | ugh, we have doc issues in ironic | 21:41 |
| TheJulia | and also 10 RLock(s) were not greened, to fix this error make sure you run eventlet.monkey_patch() before importing any other modules. | 21:41 |
| TheJulia | is getting generated through autodoc's run | 21:41 |
| JayF | I can point brain in this direction after walking the dog; I suspect we're going to have to somehow monkey_patch() earlier *or* avoid monkey_patch in the sphinx/docs case, somehow(?) | 21:42 |
| TheJulia | ... dunno | 21:43 |
| TheJulia | It fails in inspectors docs, ironic is configured the same, but we have some different errors I'm trying to chase down now | 21:43 |
| TheJulia | autodoc: failed to import module 'nics' from module 'ironic.drivers.modules.inspector'; the following exception was raised: | 21:43 |
| TheJulia | No module named 'ironic.drivers.modules.inspector.nics' | 21:43 |
| JayF | TheJulia: so despite that being logged at error, it's nonfatal: https://github.com/eventlet/eventlet/blob/master/eventlet/patcher.py#L412 | 21:45 |
| JayF | and I can confirm it was made a logged error because essentially greening locks after the fact never worked | 21:45 |
| TheJulia | yeah, ironic's gate runs without -W | 21:45 |
| TheJulia | inspector's does | 21:45 |
| JayF | this was the root cause for ironic-unit-test-crap (we monkey_patched late + eventlet couldn't green the locks) | 21:45 |
| TheJulia | just getting to other errors locall with ironic when I toggle the flag | 21:45 |
| JayF | makes sense, I just was making sure that eventlet error wasn't fatal, logging and going on is an OK way to handle that | 21:46 |
| * JayF brb need to walk the dog before too late | 21:46 | |
| TheJulia | that sounds ominous | 21:46 |
| TheJulia | somehow, autodoc is hellbent on assuming a ironic/drivers/modules/inspector/nics.py exists | 21:47 |
| TheJulia | rm -r -f .tox | 21:48 |
| TheJulia | okay, my errors on ironic were just cached data | 21:55 |
| TheJulia | from other/older runs of autodoc | 21:55 |
| TheJulia | I still get the lock error as autodoc comes along and opens the file | 21:56 |
| TheJulia | so I think it is more werkzeug | 22:06 |
| opendevreview | Merged openstack/ironic stable/2023.2: Revert "Revert "RBAC: Fix allocation check"" to use Unauthorized https://review.opendev.org/c/openstack/ironic/+/905901 | 22:06 |
| opendevreview | Julia Kreger proposed openstack/ironic-inspector master: Fix monkey_patch in autodoc execution https://review.opendev.org/c/openstack/ironic-inspector/+/906011 | 22:20 |
| opendevreview | Julia Kreger proposed openstack/ironic-inspector master: Fix monkey_patch in autodoc execution https://review.opendev.org/c/openstack/ironic-inspector/+/906011 | 22:20 |
| TheJulia | that feels slightly dirty, but it works | 22:20 |
| iurygregory | try except pass :D | 22:35 |
| iurygregory | only in inspector the problem? | 22:35 |
| TheJulia | for us afaik | 22:36 |
| iurygregory | ack | 22:36 |
| JayF | TheJulia: is this right re: https://github.com/eventlet/eventlet/issues/896 > There are cases, demonstrated by CI for https://review.opendev.org/c/openstack/ironic-inspector/+/906011/, where that isn't possible. In this specific case, Sphinx autodocs -- I'm not sure for scenarios like this how to sanely monkey_patch beforehand. | 22:38 |
| JayF | I would strongly prefer us find a better solution than 906011 though. I might spend a little time on this. | 22:38 |
| TheJulia | yeah, it doesn't feel "right", but a very similar thing was done with sphinx in general | 22:39 |
| TheJulia | locally, fwiw, I monkey patched sphinx, but the problem is it is a runner | 22:39 |
| TheJulia | at least, I think it is a separate runner process | 22:40 |
| JayF | Yeah, I think this is actually a reasonable case to ask eventlet to handle upstream or at least give an escape hatch for | 22:40 |
| JayF | I was going to take the approach of seeing if there's some sort of special variable set when doc's generation is running and if that variable is set not patching at all | 22:41 |
| TheJulia | yeah, already went looking through autodoc but it doesn't give us a firm handle of any sort | 22:41 |
| JayF | You sure there's not even something in like actual environment? At a minimum, perhaps it passes through the external environment and we could set our own canary | 22:42 |
| TheJulia | https://github.com/openstack/ironic/blob/master/doc/source/conf.py#L19-L24 <-- the prior art which I tried appling but not sure it was actually patched | 22:42 |
| JayF | Either way I'll poke at it. I've done more with Sphinx than I care to admit | 22:42 |
| TheJulia | well, I can dupe it in CI and I can dupe it locally on my desktop | 22:42 |
| JayF | Yeah I'm pretty sure I won't have any trouble duplicating it. I'm still wrapping up from walking the dog but I should be able to go fully heads down on this in like 15 | 22:43 |
| TheJulia | k | 22:44 |
| JayF | it happily duplicates lmao | 23:34 |
| JayF | and I can get around autodocs, then it just angers other stuff | 23:34 |
| JayF | (like oslo config generation) https://www.sphinx-doc.org/en/master/usage/extensions/autodoc.html#confval-autodoc_mock_imports | 23:34 |
| JayF | heh, and I just realized: importing *anything* to determine if we're doing docs generation before monkey patching eventlet will break the non-monkey_patched case | 23:37 |
| JayF | heh TheJulia I have a solution IDK if you're going to like it, but I think it's better than 906011 | 23:40 |
| TheJulia | propose it! | 23:40 |
| opendevreview | Jay Faulkner proposed openstack/ironic-inspector master: Monkey patch eventlet at base module level https://review.opendev.org/c/openstack/ironic-inspector/+/906012 | 23:41 |
| JayF | I suspect this would be the Itamar-approved fix to this bug too, I sorta asked myself what he would do with zero openstack context, it was this, and it fixed it. | 23:42 |
| TheJulia | That should fix it | 23:43 |
| JayF | it did locally. | 23:43 |
| TheJulia | ... I thought I looked for that actually and I thought it was there | 23:51 |
| TheJulia | I guess i mixed that up | 23:51 |
| JayF | it is in Ironic | 23:52 |
| JayF | which is the other clue that pointed me there | 23:52 |
| JayF | sometimes fresh eyes :D | 23:52 |
| TheJulia | ++ | 23:52 |
| JayF | that's why there are N of us :D | 23:52 |
| JayF | did you see my hilarious docs patch? | 23:53 |
| JayF | https://review.opendev.org/c/openstack/ironic/+/906001 | 23:53 |
| JayF | What a great refactor of that doc, made it 100x more readable at the expense of removing /the key piece of information from the doc/ | 23:53 |
| JayF | d'oh | 23:53 |
| TheJulia | heh | 23:53 |
| TheJulia | c'est la vie | 23:54 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!