rpittau | good morning ironic! o/ | 06:38 |
---|---|---|
arne_wiebalck | Good morning rpittau and Ironic! | 06:44 |
rpittau | hey arne_wiebalck :) | 06:44 |
kamlesh6808c | Good morning Ironic ! | 07:23 |
rpittau | iurygregory, TheJulia, about https://review.opendev.org/c/openstack/metalsmith/+/835749 we probably have to adjust the domain xml to make latest CS9 work -> https://bugzilla.redhat.com/show_bug.cgi?id=1834714 | 07:29 |
kamlesh6808c | TheJulia, iurygregory, dtantsur : This is regarding raid cleaning implementation for physical baremetal in ironic tempest plugin. | 07:58 |
kamlesh6808c | As suggested by community, i was looking to leverage introspection to grab "by_path" info for root device hint inside raid cleaning test cases. | 07:59 |
kamlesh6808c | Hurdle what I get is, while running introspection from raid cleaning block it expects to have Nova service disabled since its standalone test. If disabled, it doesn't execute inband introspection since it requires compute Service Enabled.In both sense, it skips execution. | 08:00 |
kamlesh6808c | Can you please suggest way ahead. | 08:01 |
cbouchar | TheJulia: I've been wanting to chat with you about upstream testing so let me know when available. | 10:07 |
opendevreview | Riccardo Pittau proposed openstack/ironic master: [DNM] TEST CI https://review.opendev.org/c/openstack/ironic/+/835960 | 11:08 |
rpittau | seems like grenade is not behaving very well | 11:10 |
rpittau | interesting, we're not collecting wheels for libvirt-python anymore in bifrost, that's why upgrade is failing now | 11:27 |
iurygregory | good morning Ironic | 11:41 |
iurygregory | rpittau, ack re-adjust the xml domain | 11:41 |
iurygregory | kamlesh6808c, I will check the plugin code to see if we can avoid skipping | 11:43 |
kamlesh6808c | iurygregory : thanks :) | 11:47 |
opendevreview | Riccardo Pittau proposed openstack/bifrost master: Force install binaries for libvirt-python https://review.opendev.org/c/openstack/bifrost/+/835965 | 11:48 |
rpittau | mmm may need to drop this ^ to yoga first | 11:48 |
iurygregory | hummm funny that we are always trying to fix the libvirt-python :D | 11:50 |
iurygregory | let's see how CI will react | 11:51 |
iurygregory | I will review things after the live episode | 11:51 |
rpittau | well it keeps changing from wheel to source instead of just staying to wheel :/ | 11:51 |
iurygregory | =/ | 11:53 |
iurygregory | OpenInfra Live - OpenStack Yoga: New Features Straight from the Contributors will start in 46min https://youtu.be/hwPfjvshxOM \o/ | 13:14 |
rpittau | yep grenade is utterly and completely broken | 13:15 |
rpittau | -> https://review.opendev.org/c/openstack/ironic/+/835960 | 13:15 |
TheJulia | what is grenade doing? | 13:41 |
TheJulia | also good morning | 13:41 |
iurygregory | good morning TheJulia | 13:42 |
rpittau | TheJulia: something odd with nova | 13:42 |
rpittau | or well nova/neutron combination | 13:42 |
rpittau | TheJulia: https://storage.bhs.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_c5c/835960/1/check/ironic-grenade/c5c7ca9/controller/logs/grenade.sh_log.txt | 13:43 |
iurygregory | nova + neutron </3 | 13:47 |
iurygregory | I will take a look after the presentation | 13:47 |
TheJulia | so neutron has lost it's mind I guess | 13:48 |
rpittau | it was working fine until 2 days ago | 13:49 |
TheJulia | right after the "release" | 13:50 |
rpittau | yep | 13:50 |
opendevreview | Riccardo Pittau proposed openstack/bifrost master: Force install binaries for libvirt-python https://review.opendev.org/c/openstack/bifrost/+/835965 | 13:53 |
rpittau | not related to grenade, how come we're not getting any more whl pkgs for libvirt-python ? | 13:53 |
TheJulia | I think being a binary module... it likely has to come from distros given the binary nature (that and if you ever look at how some of the translation layer stuff is built you'll want to run for the hills( | 14:04 |
TheJulia | s/hills(/hills)/ | 14:04 |
opendevreview | Riccardo Pittau proposed openstack/bifrost master: [DNM] TEST CI https://review.opendev.org/c/openstack/bifrost/+/836024 | 14:07 |
TheJulia | Well, I know why I've preferred soy milk so much recently. The coffee I've been using is slightly bitter | 14:07 |
rpittau | we were getting the whl since a couple of days ago, now they just disappeared..... | 14:08 |
TheJulia | weird | 14:09 |
* iurygregory should have added more features to the slides =( | 14:16 | |
* iurygregory will mention that we have the feedback form <3 | 14:24 | |
cbouchar | Ironic Folks: Wondering in your tempest tests whether you have hardware in the testbed with architectures other than x86_64. If so, which ones? | 14:39 |
iurygregory | I've only run tempest in x86_64 | 14:40 |
cbouchar | Do you know if there is other architecture hosts in what's set-up? | 14:41 |
ameya49 | Hi iurygregory o/ | 14:42 |
iurygregory | cbouchar, I think our CI only runs tempest in x86_64, we do run unit tests in ARM | 14:43 |
iurygregory | ameya49, hi | 14:43 |
ameya49 | Here is the tracback for quota exceeded issue while running test_introspection_basic tempest test case for ironic_tempest_plugin : https://paste.openstack.org/show/bFzCgezgUB97OP2b67Qk/ | 14:44 |
ameya49 | I was not able to add whole traceback since it was giving spam error. But let me know if you need whole traceback. Will try to upload. | 14:46 |
iurygregory | that should be enough, I will try to check latter today some codes | 14:47 |
ameya49 | sure. Thanks.. Good day ahead! | 14:47 |
iurygregory | arne_wiebalck, let us see if we will get more answers in the form after the live :D | 15:03 |
TheJulia | cbouchar: upstream all of our tests are x86_64 because virt in virt is painful and slow and we're running tests in ?9? or so public clouds. | 15:18 |
TheJulia | cbouchar: I think someone has run tempest against power, really the fundimentals are basically the same as long as the enrolled machine is flagged properly as being that architecture | 15:19 |
cbouchar | TheJulia: How do you verify the target host was reimaged successfully by last ironic action as opposed to reimage failed and booting what was on disk previously? Do you store info on the machine to verify against? | 15:25 |
TheJulia | cbouchar: two step process. We move a vm into cleaning which is a pass/fail operation | 15:32 |
TheJulia | then we redeploy | 15:32 |
cbouchar | TY | 15:32 |
TheJulia | rebuild is just expected to work as it is just the re-running deploy, fwiw | 15:33 |
cbouchar | wasn't thinking you were using VMs. | 15:33 |
TheJulia | I'm not sure that is really tested, but it is literally just an alias in our code | 15:33 |
cbouchar | TY | 15:33 |
TheJulia | yeah, all our test resources are in public clouds, so... yeah :\ | 15:33 |
rpittau | good night! o/ | 15:43 |
TheJulia | cbouchar: I have a 1.5 hour call coming up in a few minutes, and then I have an an hour of free time before my weekly 1-1 | 16:27 |
arne_wiebalck | iurygregory: not yet :) | 16:40 |
arne_wiebalck | bye everyone o/ | 16:40 |
iurygregory | arne_wiebalck, bye o/ | 16:43 |
cbouchar | TheJulia: Sounds good. | 17:04 |
* TheJulia feels exhausted | 18:22 | |
TheJulia | JayF: I think we'd document it in the admin guide, fwiw | 18:26 |
TheJulia | but yeah, afaik there is no other http medium to grok something is a folder since it is more a human construct | 18:26 |
JayF | It's one of those things where like, I don't *think* anyone would do that | 18:26 |
TheJulia | yeah | 18:26 |
JayF | but I personally know that it's possible because the static site gen I use for jay.jvf.cc and oldos.org both have it as a possibility | 18:26 |
JayF | where you can do domain.tld/path/to/file/ and it loads that as file.html | 18:27 |
TheJulia | internally is that a redirect though? | 18:28 |
JayF | No. | 18:28 |
JayF | It just works that way in browsers these days. I don't fully understand it. Logs show fetching of file.html, URL in address bar is /path/to/file/ | 18:28 |
TheJulia | funky | 18:29 |
* TheJulia momentarilly longs for HTTP and browsers of... 15 years ago | 18:29 | |
TheJulia | w/r/t https://review.opendev.org/c/openstack/ironic/+/834709/5/ironic/conductor/utils.py what specifically were you thinking config option wise | 18:30 |
JayF | It sounded like you were debating whether or not to write back image type to [the node? the image service?] | 18:34 |
JayF | and there was a little bit of disagreement | 18:34 |
JayF | I was just saying, you can make that optional | 18:34 |
TheJulia | Yeah, I'm not sure *why* we're doing it at all | 18:36 |
TheJulia | To be honest. | 18:36 |
TheJulia | That is likely Dmitry context I'll need to extract next week | 18:36 |
TheJulia | oh why oh why is grenade now failing | 18:45 |
cbouchar | TheJulia: I upgraded y bifrost host to yoga (with adjustments as needed) and applied https://review.opendev.org/c/openstack/ironic/+/834709. It worked like a charm. | 18:52 |
TheJulia | woot! | 19:28 |
TheJulia | cbouchar: that makes me super happy to hear | 19:30 |
TheJulia | so before 2 days ago, grenade was Xena->Master, now it is Yoga->Master and it *feels* like filesystem permissions are wrong | 20:00 |
TheJulia | file permissions seem suspect | 20:13 |
TheJulia | but odds it broke now | 20:14 |
* TheJulia goes and lays down | 20:15 | |
admiyo | Containers kinda suck, don't they? | 20:43 |
admiyo | I mean...they let you do some cool things, but once they start failing and hiding information on you, they really are nasty little beasts. | 20:44 |
admiyo | I mean, it does tell me the pid file has a number in it that does not match, but that really is not actionable data | 20:44 |
admiyo | (No such process)? Reaslly...why did it die? "I'm sorry Adam, I can't tell you that." | 20:45 |
admiyo | Is OVS like, a major thorn in the Ironic paw? | 20:45 |
admiyo | JayF, um, that is a dangerous mechanism, and is severly limited in the browser. Think XSRF type attack. | 21:01 |
admiyo | I mean, yeah, I can do file:/// but only if I either explicitly type it in or from another page already loaded file file:/// | 21:02 |
admiyo | There is an apache mod that lets you od http://tld/~user/ and that will load from the users public_html dir, but there are a slew of permission issues there now. That was also all way before SELinux got on the scene | 21:03 |
JayF | admiyo: I'm not saying it's good; I'm saying we should explicitly document that we don't support it since it's possible. | 21:17 |
JayF | Also, all my comments on PRs these days are mostly advisory, since I haven't been working full time on Ironic in several months at this point | 21:18 |
admiyo | Oh, I am just speaking off the cuff here, just spent a lot of time dealing with some security stuff like that | 21:19 |
admiyo | I'm still trying to understand the issue | 21:20 |
JayF | It's not an issue. | 21:21 |
JayF | A PR TheJulia put up has an assumption that any URL ending in / is a path, not a single file | 21:22 |
admiyo | Cool | 21:22 |
JayF | which is an OK assumption for most Ironic use cases, I just think we should doc it | 21:22 |
admiyo | Well...I don't think there is anything in the HTTP spec that distinguishes that. I know a lot of autogenerated content is actually kinda sloppy about differentiating. I think the most common approach now is to do a redirect | 21:24 |
admiyo | For Keystone I am pretty sure the Flask mechanism treats them the same. I know we don't have seprate entries for /v3 and /v3/ and they both return the same data | 21:27 |
admiyo | but you are talking images. Really it should be based on the content type of the response, not the path. | 21:28 |
* admiyo turning back into a purist, goes back to debuggin OVS | 21:28 | |
TheJulia | admiyo: yeah, capturing redirects now too. | 21:32 |
opendevreview | Iury Gregory Melo Ferreira proposed openstack/ironic-python-agent-builder master: Add multipath element https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/836094 | 21:37 |
admiyo | TheJulia, I am trying to stifle the pedant in me, but...you should make no assumptions about the URL except based on the content type it returns. But I am guessing here the issue is just that it says "binary" for most images. Anyd you want to make a decision before you actually call on the URL.... | 21:38 |
TheJulia | Yeah, non-images actually. Like to a repository path used by anaconda | 21:41 |
JayF | What really matters in this case is that we capture the 99.9% case and document that we don't support the 0.1% case (probably need more nines and 0s in those comparisons) | 22:02 |
TheJulia | ++ | 22:18 |
opendevreview | Julia Kreger proposed openstack/ironic stable/yoga: DNM: get some additional info to debug grenade https://review.opendev.org/c/openstack/ironic/+/836105 | 23:18 |
opendevreview | Julia Kreger proposed openstack/ironic master: DNM: get some additional info to debug grenade https://review.opendev.org/c/openstack/ironic/+/836106 | 23:19 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!