| opendevreview | Takashi Kajinami proposed openstack/ironic-lib stable/2024.2: Replace outdated python job template https://review.opendev.org/c/openstack/ironic-lib/+/933898 | 00:15 |
|---|---|---|
| zigo | Hi there! This isn't the first time for me: test_format_inspector.TestFormatInspectors.test_iso_udf times-out and then my Ironic package is declared unbuildable. | 08:38 |
| zigo | Does one know where to control the timeout value for this test? I've looked for it, but didn't find where. | 08:38 |
| masghar | zigo: no idea | 11:07 |
| masghar | https://www.irccloud.com/pastebin/IHCN1bu1/ | 11:11 |
| masghar | Oh whoa how did my message turn into a pastebin xD | 11:12 |
| masghar | FWIW the tests/scenario directory does exist in https://github.com/openstack/ironic-tempest-plugin | 11:19 |
| iurygregory | this is weird | 11:23 |
| iurygregory | to me it seems tempest didn't find our plugin, but it was installed from what I saw in the logs | 11:23 |
| iurygregory | maybe something changed in tempest :thinking: | 11:24 |
| masghar | Yeah it doesnt seem like a flake either, somehow - I think if we retest it will happen again | 11:24 |
| masghar | Perhaps, maybe its using a different branch? | 11:25 |
| iurygregory | it failed on both jobs same way... | 11:25 |
| masghar | Yes | 11:25 |
| iurygregory | let me check something here | 11:25 |
| masghar | Well we record quite a lot of logs, don't we 😅 | 11:39 |
| masghar | Maybe it is a flake? https://zuul.opendev.org/t/openstack/builds?job_name=ironic-lib-bios-ipmi-src&project=openstack%2Fironic-lib&skip=0 | 11:43 |
| iurygregory | I have the feeling is a flake | 11:50 |
| iurygregory | but trying to check if tempest changed something | 11:50 |
| iurygregory | ohhh we are using ubuntu-jammy, I remember there was a problem after we dropped py38 | 11:53 |
| iurygregory | maybe is related | 11:53 |
| iurygregory | we need to switch the master to the new ubuntu I think | 11:53 |
| iurygregory | gmann, do you think this is the case? we saw some strange issues with tempest on master... | 11:54 |
| masghar | I see! Perhaps it could be related | 12:13 |
| iurygregory | one way we can probably check is doing a recheck in stable job that also failed | 12:26 |
| masghar | Lets recheck and se | 12:26 |
| masghar | see* | 12:26 |
| iurygregory | but the failure was today... | 12:26 |
| iurygregory | let me see the thread in the ML | 12:27 |
| iurygregory | oh ok, the thread was [qa][all] stable/2023.1 gate is blocked (hold the recheck) | 12:27 |
| iurygregory | so maybe is another issue we have | 12:28 |
| masghar | Yeah | 12:28 |
| masghar | I'm trying to find the where ironic-base is: https://github.dev/openstack/ironic-lib/blob/467e7ecc18f65985cbd833687b85b717ad363183/zuul.d/ironic-lib-jobs.yaml#L3-L4 | 12:29 |
| iurygregory | in openstack/ironic =) | 12:29 |
| masghar | oh I see! | 12:29 |
| TheJulia | Good morning | 13:16 |
| JayF | zigo: it's a test that uses a lot of io and has failed similarly in our CI. I suggest patching it out. Just flaky | 13:18 |
| zigo | JayF: I had the same issue with other projects... | 13:19 |
| TheJulia | Sorry for not being around yesterday... I woke up, remembered to check if I had jury duty... | 13:21 |
| TheJulia | and you can imagine me running out the door heading towards the courthouse | 13:22 |
| zigo | JayF: Is there anyone working on https://bugs.debian.org/1084566 ? | 13:24 |
| iurygregory | same failure D: | 13:45 |
| JayF | Zigo: There's no effort being made right now on python 3.13 compatibility. I suspect if you file an issue with us, someone will eventually get around to it, or you're welcome to submit a patch. | 14:12 |
| JayF | But that being said, I would not assume that everything is going to behave hunky dory on python 3.13 unless you do full integration testing, just because I'm sure they're going to be more eventlet shenanigans | 14:12 |
| masghar | iurygregory: controller/logs/devstacklog.txt does say 'Successfully installed ironic-tempest-plugin-2.11.1.dev8' | 14:17 |
| opendevreview | cid proposed openstack/ironic master: Save ``configdrive`` in an auxiliary table https://review.opendev.org/c/openstack/ironic/+/933622 | 14:39 |
| masghar | I am comparing a successful run (https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_903/933029/2/check/ironic-lib-uefi-ipmi-src/903d952/job-output.txt) with the recent failed run | 14:43 |
| masghar | iurygregory: If you put them side by side, and search for 'TASK [run-tempest : Run Tempest]', you will see that the successful run was using setuptools (which is deprecated) | 14:44 |
| masghar | The new run has lines like 'controller | all recreate', 'controller | all installdeps:' instead | 14:46 |
| opendevreview | Thomas Goirand proposed openstack/ironic master: Remove the use of crypt.crypt, removed in Python 3.13 https://review.opendev.org/c/openstack/ironic/+/933917 | 14:54 |
| zigo | Is this done right? ^ | 14:55 |
| zigo | JayF: Yeah, I'm worried with the Eventlet in 3.13. I need this fixed by February (date of Trixie freeze...)... | 14:58 |
| opendevreview | cid proposed openstack/ironic master: Save ``configdrive`` in an auxiliary table https://review.opendev.org/c/openstack/ironic/+/933622 | 15:13 |
| opendevreview | Merged openstack/sushy-tools master: Remove Python 3.8 support https://review.opendev.org/c/openstack/sushy-tools/+/933259 | 15:18 |
| TheJulia | JayF: replied to some of your comments on https://review.opendev.org/c/openstack/ironic-specs/+/933612 to try and bring a bit more clarity. Adam from metal3 or even cardoe might have additional thoughts to help us clarify overall vision. | 15:22 |
| cardoe | bah yes I need to do that thank you for reminding me. I'm over here chasing yaks and shaving squirrels. | 15:23 |
| TheJulia | ... Why would you *shave* squirrels?! | 15:28 |
| TheJulia | clearly you need a fleet of cats. | 15:28 |
| opendevreview | Thomas Goirand proposed openstack/ironic master: Remove the use of crypt.crypt, removed in Python 3.13 https://review.opendev.org/c/openstack/ironic/+/933917 | 15:30 |
| TheJulia | Or Corgis. | 15:30 |
| TheJulia | Granted, the Corgi's will jsut keep the Squirrels away, or cuddle them. It remains to be seen exactly what one would do | 15:30 |
| opendevreview | Verification of a change to openstack/ironic-python-agent master failed: Correct invalid docstrings; s/Found/Error/ https://review.opendev.org/c/openstack/ironic-python-agent/+/911598 | 15:38 |
| cardoe | I knew I was doing something wrong. | 15:38 |
| cardoe | I also need to reply to you on the ML. | 15:38 |
| cardoe | Something is up with tempest and IPA | 15:39 |
| TheJulia | So, any further though on graphical consoles? The more I think about it, the more I want to just replace console == serial console | 15:49 |
| TheJulia | cardoe: got a job link? | 15:49 |
| cardoe | https://zuul.opendev.org/t/openstack/build/f0e3e3dc4819477bbb5e40b141aa595f is one | 15:50 |
| cardoe | https://zuul.opendev.org/t/openstack/build/32fdc8fa2ac24b64b92babae105a2594 is another | 15:50 |
| cardoe | when you say "console == serial console". We're saying that the word "console" will mean serial console always? | 15:50 |
| cardoe | or just removing the ambiguous use of "console" and always prefixing it with "serial" or "graphical"? | 15:51 |
| TheJulia | today it does | 15:51 |
| TheJulia | And shellinabox is dead... sooooo | 15:51 |
| TheJulia | maybe best to just replace the idea/meaning to be more current with graphical | 15:51 |
| TheJulia | we're also dealing with a barely used 10 year old concept | 15:51 |
| TheJulia | https://f3d5be211d88562b5165-96f941636ebce386b2873cc66e81b34b.ssl.cf5.rackcdn.com/927544/5/check/ironic-tempest-ipa-partition-uefi-pxe-grub2/f0e3e3d/controller/logs/ironic-bm-logs/node-0_console_2024-10-30-19%3A22%3A11_log.txt <-- trying to uefi boot cirros is known bad idea | 15:55 |
| TheJulia | we need to fix that | 15:55 |
| TheJulia | that second one is *weird* | 16:00 |
| TheJulia | The second one just appears to fall over from tempest? Did we change a job name?! | 16:06 |
| TheJulia | Well, a test name | 16:06 |
| TheJulia | it never actually tries to deploy a node | 16:06 |
| TheJulia | cardoe: I guess instead of do more interface work, just move past "serial" | 16:13 |
| TheJulia | Has anyone heard if HPE Gen11 gear, specifically with AMD processors, takes a long time to reboot? | 16:17 |
| TheJulia | Nevermind... https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00128973en_us | 16:18 |
| JayF | TheJulia: lol | 17:15 |
| JayF | TheJulia: I personally saw one take ~30 minutes | 17:15 |
| JayF | there's a reason we want to enable OOB steps for OEM-exposed things in redfish, we're literally paying ^^ that amount of time each cleaning/deployment to do ESKM stuff | 17:15 |
| TheJulia | eek, yeah | 17:18 |
| JayF | rpittau: are you working on a priorities spec for this ptg? | 17:20 |
| JayF | rpittau: I was aobut to do prep for a quick video on it, and realized that prep is like, 50% of the work to make that doc | 17:20 |
| JayF | I might draft something and put it up | 17:21 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Implement basic interfaces for GraphicalConsole Interface https://review.opendev.org/c/openstack/ironic/+/547356 | 17:22 |
| TheJulia | so ^ makes me feel like a whole interface to keep serial is just... overkill | 17:22 |
| TheJulia | when maybe serial should be stamped out | 17:22 |
| TheJulia | by the robo-feature-eliminator-5000 | 17:22 |
| JayF | I am +1 to retiring serial, but less +1 to reusing the same interface names for the new stuff | 17:22 |
| JayF | I think we deprecated shellinabox when I was PTL | 17:23 |
| TheJulia | I'm sort of entirely against putting in a new interface | 17:23 |
| JayF | or at least decided to, sometimes I missed the follow thru one some things (/me looks at that python-ironicclient patch for shards) | 17:23 |
| TheJulia | which is a huge shift for me | 17:23 |
| JayF | What value, other than less code churn, does reusing the same interface have? | 17:23 |
| TheJulia | similar meaning, we can reuse the same exact webapi interface | 17:24 |
| TheJulia | which we would realistically *need* to do to keep it simple as it sort of does exactly what is needed to day anyhow | 17:24 |
| TheJulia | just the driver under-the-hood is a different flow | 17:24 |
| JayF | ooooh, so like, we avoid client work, api ref updates, etc | 17:24 |
| TheJulia | bbiingo | 17:24 |
| JayF | and shellinabox was already "we give you a link to click" style, yeah? | 17:24 |
| TheJulia | new field, we begin cutting a harder path for ourselves | 17:24 |
| TheJulia | sort of like "we can take the path already paved and do some side work" or "we must bore through this mountain to walk forward" | 17:25 |
| JayF | honestly I just got whooshed for a second that interfaces are API-facing | 17:25 |
| cardoe | I'm with Julia on depreciate and reuse. | 17:25 |
| TheJulia | shellinabox was sort of that path as well | 17:25 |
| TheJulia | I don't remember the exact internals | 17:25 |
| cardoe | If I understand the old interface correctly. Which I think was just "here's a link to click" | 17:25 |
| TheJulia | but I know hpe had a patch for ages which returned a vnc console url for nova to proxy through to | 17:25 |
| cardoe | and it took you to a shellinabox web page thing. | 17:25 |
| JayF | yeah, which is what nova does, and what we'll probably do w/a vnc url I assume | 17:26 |
| TheJulia | yeah | 17:26 |
| JayF | yep, I'm completely sold now already | 17:26 |
| cardoe | Yeaj | 17:26 |
| TheJulia | cool cool, we should like solidfy this consensus with concrete | 17:26 |
| JayF | it seems more like the wrong decision was delineating it as **serial console** originally | 17:26 |
| TheJulia | yeah | 17:26 |
| TheJulia | Well, I think the argument was "people want to use both" | 17:26 |
| TheJulia | and the more cases I've gotten pulled into with serial console use tanking host performance | 17:27 |
| JayF | I've never met a person who had that as a requirement tb h | 17:27 |
| TheJulia | the more I'm just "no, don't do that" | 17:27 |
| JayF | ++ | 17:27 |
| JayF | it's old technology grafted on, like the CSM modules in UEFI firmwares | 17:27 |
| TheJulia | yeah | 17:27 |
| TheJulia | and even with UEFI, the host is graphical out of the box anyway | 17:27 |
| JayF | I still see this as like, kinda separate? We can retire shellinabox right now if we want, and just leave no implementations, and fill this one in as we go | 17:27 |
| TheJulia | its not like it is a serial port on a Vax | 17:27 |
| JayF | yep | 17:27 |
| TheJulia | yup, exactly | 17:28 |
| TheJulia | and just build a new redfish interface with the backend | 17:28 |
| JayF | only real update that'd need to happen is probably in docs | 17:28 |
| TheJulia | a whole/complete/secure backend | 17:28 |
| JayF | just s/serial// in places | 17:28 |
| TheJulia | I'm not even sure that will entirely be needed because we just famed it all as "console" by default | 17:28 |
| TheJulia | but yeah, we'll need to check | 17:28 |
| TheJulia | I'll go start a spec revision shortly | 17:28 |
| JayF | sgtm | 17:29 |
| JayF | also I responded on container spec | 17:29 |
| JayF | tl;dr: "we can just trust podman for that part" sounds a lot like "we can just trust qemu-img for that part" | 17:29 |
| JayF | which means I'd think it's slightly wise to ensure they /actually are expecting use with untrusted images/ | 17:29 |
| TheJulia | yes, but what is k8s doing at that point since contents are inherently untrustedf | 17:30 |
| JayF | and maybe even document that decision so if later they try to takesies backsies we got evidence LOL | 17:30 |
| JayF | I don't understand what you mean with your question? | 17:30 |
| JayF | if there was a bug, where I could somehow craft an OCI image that when executed by podman would ... say put the contents of /etc/passwd into that container instead | 17:30 |
| JayF | would they treat it as security or just tell us to stop using untrusted images | 17:31 |
| TheJulia | so I guess the issue in my mind is we use a tool like podman to do the underlying retrieval of the container out of the box and then the cracking open of it. It is not like we download it and then extract it | 17:33 |
| TheJulia | so if there *is* a problem with the contents of the container, everyone else using the same tooling is impacted | 17:33 |
| JayF | that was all true for our qemu-img bugs | 17:33 |
| TheJulia | no, we knew there issues and had code to guard against some of them | 17:34 |
| TheJulia | but not all of them and we didn't knew the base issue was that bad | 17:34 |
| JayF | Yes; but everyone using that same tooling was impacted | 17:34 |
| TheJulia | otherwise, were creating our own retrieval and extraction code in python | 17:34 |
| TheJulia | which means, we;re sort of building a mountain of work | 17:34 |
| TheJulia | Indeed, that sometimes happens, unfortunately | 17:34 |
| JayF | I'm not saying "don't trust podman", I'm saying it'd be extremely nice to point to a doc that indicates they have a security policy that matches our expectation | 17:34 |
| TheJulia | AHH | 17:35 |
| TheJulia | okay | 17:35 |
| JayF | because I no longer trust obvious assumptions about software we use | 17:35 |
| TheJulia | That, I think makes a ton of sense | 17:35 |
| TheJulia | Reasonable | 17:35 |
| JayF | where if a qemu-img style bug came out, instead of us writing python to detect bad containers, the oci/podman/docker ecosystem cares enough to fix it at that level | 17:35 |
| shermanm | on the topic of deprecating serial console and replacing with graphical: is this assuming that all supported hardware would expose a graphical kvm interface of some kind? | 17:35 |
| JayF | shermanm: the only interface currently implemented for serial is using an EOL, insecure project. The only proposed *new* interface is redfish, which AIUI is specified to be VNC | 17:36 |
| TheJulia | shermanm: the assumption is some can, not everything *can* | 17:36 |
| JayF | I don't think we'd -1 another serial implementation in this interface, but I also doubt it'll make sense for server hardware | 17:36 |
| TheJulia | well, kvmip which is just vnc with some byte structures flipped | 17:36 |
| TheJulia | That is my feeling as well | 17:36 |
| TheJulia | w/r/t if someone had a redfish raw serial interface | 17:37 |
| shermanm | I ask because we have a hard requirement to have some kind of user-facing console access for a pretty wide range of hardware variants, some of which don't support redfish at all, some of which still only have terrible java kvms, etc. | 17:37 |
| shermanm | doesn't need to be a serial console, but a fair chunk of our hardware won't necessarily play nice with graphical consoles directly | 17:37 |
| JayF | That's a borderline-impossible requirement to do with the resources I assume you have | 17:37 |
| JayF | at least completely out of band | 17:38 |
| JayF | Hell last time I tried to use one of those old java-based consoles on a drac, I think I had to break out a VM with IE6 inside | 17:38 |
| JayF | or something similarly ridiculous | 17:38 |
| TheJulia | shermanm: the secret is, they don't with serial *either* :) | 17:38 |
| shermanm | well, right now we're using the serial console ipmitool sol just fine, but it sounds like that's depending on an EoL project in ironic? | 17:38 |
| JayF | shermanm: if you're using the shellinabox implementation, and exposing it to customers; tldr AHHHHHHHHHH | 17:38 |
| TheJulia | shermanm: so, there are two ways under the hood in ironic today | 17:39 |
| TheJulia | shellinthebox and an awful ipmitool sol proxy driver | 17:39 |
| JayF | https://github.com/shellinabox/shellinabox (last commit: 6 years ago) | 17:39 |
| shermanm | it's not great, but it's core functionality for us, and we've got a not completely untrusted set of users | 17:39 |
| TheJulia | err, shellinabox | 17:39 |
| TheJulia | the ipmitool socat console just gives you a port you telnet to... | 17:39 |
| TheJulia | note I said telnet, it is *awful* | 17:39 |
| JayF | Sometimes I'm glad I haven't read /all/ the code in Ironic. :) | 17:40 |
| shermanm | it is pretty bad, but it's "something" when a user builds a kernel and breaks their boot | 17:40 |
| JayF | shermanm: At Rackspace, with OnMetal, we developed rescue mode for this case | 17:40 |
| TheJulia | yeah, something is better than nothing in many cases | 17:40 |
| JayF | shermanm: so that is an alternative in cases where console may not be possible, and works more universally | 17:40 |
| TheJulia | ++, I'll note with serial port stuff today, for hardware in UEFI mode, nothing may get logged | 17:41 |
| TheJulia | at least, logged to the serial port | 17:41 |
| TheJulia | since, the default mode of user interaction outside of headless gear is the graphics card | 17:41 |
| shermanm | this does make sense, I'll need to dig into what we have, and see if anything's truly relying on serial only, and couldn't be resolved with rescue mode. I freely admit that we've got a lot of edge cases | 17:44 |
| TheJulia | yeah, we're not saying we're going to smash all things serial console with a giant mallot | 17:45 |
| TheJulia | .... As much as I kind of want to be a loony toons character with a mallot at the moment. :) | 17:46 |
| TheJulia | Just driving towards graphical supercedes the use model | 17:46 |
| TheJulia | And that is a road | 17:46 |
| TheJulia | Which we need to still drive down | 17:46 |
| TheJulia | we're in the "aruging over which route to select in google maps" phase | 17:46 |
| shermanm | yep! I just wanted to raise a hand sayin "8 year old weird hardware still kinda needs this crappy thing" | 17:51 |
| gmann | iurygregory: tempest jobs should be fixed now, please recheck | 17:52 |
| JayF | shermanm: well, the only plan is to kill off shellinabox afaict right now; we haven't deprecated the serial proxy piece, so I think you're in good shape. You shouldn't deploy shellinabox anyway :D | 17:54 |
| jrosser | I also use the ipmi/sol + serial proxy today for arm hosts | 17:56 |
| jrosser | I expect that could be more modern with redfish/ssh + serial proxy in the future | 17:57 |
| TheJulia | jrosser: this is encouraged, but I'm not sure the upstream cores can just make it happen without some assitance | 17:58 |
| shermanm | JayF: thanks for that clarification, scrolling up some more I realize I didn't have the full context. We're using the ipmitool-socat interface at the moment, the the serial proxy, not shellinabox | 17:59 |
| shermanm | I'd been considering writing a different backend using something like goconserver instead of ipmitool | 17:59 |
| TheJulia | so thinking about it | 18:06 |
| TheJulia | if we *just* return a VNC URL today, that plugs into nova's vnc proxy kind of cleanly | 18:07 |
| TheJulia | we *could* create a proxy of our own, but is it *really* needed except for standalone users?! | 18:07 |
| TheJulia | #discuss :) | 18:09 |
| shermanm | even if the proxy was out-of-tree, otherwise not recommended, having an interface we could leverage to deploy / build our own would be useful | 18:10 |
| TheJulia | I don't think that is anything we would block :) | 18:12 |
| opendevreview | Julia Kreger proposed openstack/ironic-specs master: Move graphical vnc console interface to retired status https://review.opendev.org/c/openstack/ironic-specs/+/933937 | 18:29 |
| TheJulia | okay, that ^^^ commit message is a bit verbose | 18:29 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: WIP: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 18:59 |
| JayF | shadower: almost all of ironic is pluggable, even if we don't technically advertise support for it (or promise any API compat) | 19:25 |
| JayF | er, shermanm ^^ sorry shadow | 19:25 |
| JayF | shermanm: e.g. https://opendev.org/openstack/ironic/src/branch/master/setup.cfg#L87 you can use the same HWM pattern from IPA to inject interfaces and hardware types (just remember you'll need a custom hw type if you want to declare compat with that interface) | 19:25 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 20:29 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 20:35 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 20:35 |
| JayF | anyone know WTF is up with python-ironicclient functional tests? They appear massively broken, and are nonvoting https://zuul.opendev.org/t/openstack/build/034561574cb244cf98e2cc0826187a97 | 20:50 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 20:50 |
| JayF | we should find a path to fix them OR remove them entirely imo | 20:50 |
| JayF | adamcarthur5: ^ cc | 20:51 |
| TheJulia | oh my | 20:56 |
| TheJulia | so, they use some testing framework from oslo which I bet got nuked | 20:56 |
| TheJulia | at least, that is what comes to mind when thinking about them | 20:56 |
| JayF | yeah; I think adamcarthur5 is going to try and reproduce in his local devstack and see if there's a quick path to a fix | 20:57 |
| TheJulia | ack | 20:57 |
| TheJulia | I wouldn't mind just nuking those functional tests from high orbit if it came down to it | 20:58 |
| adamcarthur5 | You guys overhauled auth right? https://opendev.org/openstack/python-ironicclient/src/branch/master/doc/source/contributor/testing.rst#functional-testing | 20:59 |
| adamcarthur5 | That could be why? | 20:59 |
| JayF | oh, the scope changes! of course! | 20:59 |
| JayF | does that line up?! | 20:59 |
| TheJulia | well, I think we made changes to fix it | 20:59 |
| TheJulia | but if the underlying object is missing.... | 20:59 |
| JayF | releasenotes/notes/change-default-rbac-policy-f2f154043910f26a.yaml was created Date: Mon Nov 27 13:46:34 2023 -0800 | 21:00 |
| JayF | timing does /not/ line up | 21:01 |
| JayF | not saying it means it's not a cause, but less likely | 21:01 |
| adamcarthur5 | Yeah but if this test has been non-voting for a while, the problems are likely stacked | 21:01 |
| JayF | ++ exactly, itamar hit that with snmp | 21:01 |
| JayF | fix vpdu; ironic snmp driver broke in the interim :( | 21:01 |
| adamcarthur5 | Julia you mentioned nuking this, Jay, how desperate are you on actually getting to the root cause of this as part of this patch? | 21:02 |
| JayF | lol | 21:02 |
| JayF | give it a swift kick in the rear on the way outta the repo :D | 21:02 |
| JayF | just make sure it's a separate commit from shards | 21:02 |
| JayF | and provide lots of context in the commit message | 21:02 |
| adamcarthur5 | Okay. In that case, I am going to investigate for 10-15 mins to maximise context I can give | 21:03 |
| JayF | ++ that sounds like a good idea, make sure it's not low hanging | 21:03 |
| JayF | thanks! | 21:03 |
| * TheJulia whispers "nuke it" | 21:03 | |
| * TheJulia thinks it is time to go to the post office | 21:04 | |
| adamcarthur5 | Yeah, it seems the current error is auth: AttributeError: 'BaremetalPortGroupTests' object has no attribute 'auth_ref' | 21:06 |
| adamcarthur5 | But that seems fixable.. I will note it down. | 21:11 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Support /v1/shards https://review.opendev.org/c/openstack/python-ironicclient/+/933894 | 21:12 |
| adamcarthur5 | https://review.opendev.org/c/openstack/python-ironicclient/+/908828... Huh, I was a reviewer on this change. It must have been from my codespell work | 21:16 |
| adamcarthur5 | Tests last past here: Over a year ago https://review.opendev.org/c/openstack/python-ironicclient/+/891560. There were only doc changes + running of older commits (some from 2017) in CI | 21:23 |
| adamcarthur5 | That might align with an auth change? | 21:26 |
| opendevreview | Adam McArthur proposed openstack/python-ironicclient master: Remove Functional Tests https://review.opendev.org/c/openstack/python-ironicclient/+/933967 | 21:33 |
| adamcarthur5 | Let me know if this is enough detail, although I do have one suspicion for a fix. Let me know | 21:33 |
| JayF | let me put it this way: they either gotta be fixed or they gotta go | 21:43 |
| JayF | The question is are they providing value | 21:43 |
| JayF | I don't know the answer to ^ that | 21:43 |
| * adamcarthur5 my suspicion was wrong | 21:47 | |
| opendevreview | Adam McArthur proposed openstack/ironic master: api: Introduce new mechanism for API versioning https://review.opendev.org/c/openstack/ironic/+/928919 | 22:22 |
| adamcarthur5 | The above was tested with https://gist.github.com/Sharpz7/97356eb57f77d3ee75892791c6cab155 | 22:24 |
| opendevreview | Adam McArthur proposed openstack/ironic master: api: Introduce new mechanism for API versioning https://review.opendev.org/c/openstack/ironic/+/928919 | 22:27 |
| adamcarthur5 | Hey folks, I am trying to add some tests to ironic-tempest-plugin, I am assuming it needs a devstack to test locally. Does it need to be a fresh devstack? Or is the repo self-cleaning? Is it the normal pattern to test locally? | 22:50 |
| opendevreview | Adam McArthur proposed openstack/ironic-tempest-plugin master: WIP: Microversion handling test https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/933970 | 22:57 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!