Tuesday, 2024-10-22

rpittaugood morning ironic! o/07:10
rpittaucardoe: yes for sushy release, definitely, we have a guide for the release if you want to take care of that https://docs.openstack.org/ironic/latest/contributor/releasing.html07:31
opendevreviewMerged openstack/ironic-tempest-plugin master: Drop unnecessary 'x' bit from doc config file  https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/93286507:51
opendevreviewMerged openstack/ironic-ui master: Drop unnecessary 'x' bit from doc config file  https://review.opendev.org/c/openstack/ironic-ui/+/93286607:52
opendevreviewVerification of a change to openstack/ironic-python-agent stable/2024.1 failed: Remove non RE2 job config  https://review.opendev.org/c/openstack/ironic-python-agent/+/93266007:58
opendevreviewMerged openstack/networking-baremetal master: Drop unnecessary 'x' bit from doc config file  https://review.opendev.org/c/openstack/networking-baremetal/+/93287207:58
dtantsurJayF: this could be a fallout from the CVE: https://bugs.launchpad.net/ironic-python-agent/+bug/208533109:06
dtantsuractually... I need to check if Metal3 does the right thing, hold on09:09
opendevreviewDmitry Tantsur proposed openstack/ironic-python-agent master: Warn when the provided checksum algorithm does not match the detected  https://review.opendev.org/c/openstack/ironic-python-agent/+/93297709:38
iurygregorygood morning11:10
iurygregory_power outage :facepalm:11:37
*** iurygregory_ is now known as iurygregory11:38
opendevreviewcid proposed openstack/ironic master: Gracefully handle bad request exception  https://review.opendev.org/c/openstack/ironic/+/93184912:02
TheJuliagood morning12:40
rpittaugood morning TheJulia :)12:55
rpittauironic PTG starting in 5 minutes! https://meetpad.opendev.org/ironic-ptg-october-202412:55
cardoethis conf has a distinct lack of coffee. staggering into the morning keynote and able to just fill up a cup on my way in of $RANDOM blend is what I've become accustomed to.12:56
dtantsurAlas!12:57
cardoeanother random thought while I wait for coffee to brew... how do flat earthers reconcile time zones?12:59
iurygregoryoh no12:59
iurygregorykkkkkkkkkkkk12:59
rpittaucardoe: there is no such thing as time zones in flat-heart-world :)13:00
TheJuliacardoe: your using logic, be careful or you'll break brains ;)14:27
cardoeheh14:28
* cid :d wait, was I stopping people from speaking?15:02
JayFno, you're good :) 15:06
* cid thought I was moderator just like yesterday (weird)15:06
opendevreviewDoug Goldstein proposed openstack/ironic master: dedup reboot request in redfish bios path  https://review.opendev.org/c/openstack/ironic/+/93302015:08
cardoerpittau: can I propose another topic?15:16
cardoePython typing15:17
rpittaucardoe: you mean today or in general?15:20
cardoein general15:20
rpittaucardoe: sure :)15:20
JayFit'd be fun to do this now :)15:23
cardoeLet's do it now then15:24
opendevreviewPavlo Shchelokovskyy proposed openstack/ironic master: Update some ipmi config option help strings  https://review.opendev.org/c/openstack/ironic/+/93302315:33
cardoeskrobul pointed out I broke something so I'll be afk from the session for a bit. ping me here if it's one of my topics.16:07
opendevreviewMerged openstack/ironic-lib master: add pyproject.toml to support pip 23.1  https://review.opendev.org/c/openstack/ironic-lib/+/93273816:07
opendevreviewDmitry Tantsur proposed openstack/ironic-lib master: Drop the unused capabilities module  https://review.opendev.org/c/openstack/ironic-lib/+/93302916:25
dtantsurJayF: ^^^16:25
dtantsurfeel free to double-check that it's not used16:25
opendevreviewMerged openstack/ironic-python-agent master: Warn when the provided checksum algorithm does not match the detected  https://review.opendev.org/c/openstack/ironic-python-agent/+/93297716:49
cardoerpittau: when ya get some time could you weigh in on https://review.opendev.org/c/openstack/sushy/+/93209616:51
JayFis https://review.opendev.org/c/openstack/ironic/+/931849 backportable? WDYT?17:07
TheJuliashould be, I think17:16
TheJuliaregardless it should have raised a bad request response17:16
opendevreviewJay Faulkner proposed openstack/ironic stable/2024.2: Gracefully handle bad request exception  https://review.opendev.org/c/openstack/ironic/+/93303617:18
opendevreviewDoug Goldstein proposed openstack/ironic master: WIP: redbios BIOS apply settings support disable_ramdisk  https://review.opendev.org/c/openstack/ironic/+/93303717:23
cardoeJayF: My little script to handle "git review" dropped me from the meeting17:24
JayFwe dropped the meeting anyway because it was outta steam :)17:24
cardoeIt turns off the VPN and then turns it back on and that blipped me17:24
JayFand I've been on for like 4.5 hours 17:24
JayFchipotle will be required :D 17:24
cardoeThat's the WIP change above I was asking about.17:24
cardoeI'll ask about it tomorrow then.17:25
JayFI have it open17:25
JayFwill look and comment async17:25
JayFitamarst, same GR-OSS python dev who helped out some with the uplift of eventlet last year, is going to take a look at the virtualpdu upgrade17:39
JayFso we'll see if he can apply his magical async python knowledge to help us out17:39
JayFcardoe: You know, I wonder if you just want two steps, not one17:44
JayFcardoe: one with a guaranteed reboot, one without17:44
JayFcardoe: maybe BIOSInterface.stage_configuration17:45
JayFwhich would make sense for "setup the bios config for next reboot" without the reboot17:45
JayFand would leave the existing behavior intact for people who want the reboot17:45
cardoeWell I'm leaving the reboot behavior intact.17:45
cardoeSo you have to reboot.17:46
JayFSo prior to that change: IPA's heartbeat was the "ok the bios config applied" (?)17:47
JayFand after that change, the hope is, just the idea that it powered off/on is enough to know the config applied17:47
JayFit's not easy code to follow (not really your fault)17:47
opendevreviewMerged openstack/ironic unmaintained/xena: Fix actual size calculation for storage fallback logic  https://review.opendev.org/c/openstack/ironic/+/93128518:08
opendevreviewMerged openstack/ironic unmaintained/wallaby: Fix actual size calculation for storage fallback logic  https://review.opendev.org/c/openstack/ironic/+/93128618:08
cardoeSo the existing code path is run the "reset BIOS" or the "apply pile of BIOS settings" step. You must be booted into the IPA image for this to work.18:08
cardoeIt hits the redfish endpoints for either of those and then reboots the machine. The BMC is responsible for seeing that someone has staged either of those two operations and then applies them before booting the machine.18:09
cardoeIt's implementation specific if you DON'T reboot after doing "reset BIOS" or "apply pile of BIOS settings". Some machines are happy to continue servicing other Redfish requests. While others give you various levels of "don't talk to me".18:10
cardoeWhile some will clear the pending request if you do something else.18:11
cardoeThe box reboots and IPA comes back up and loads back into the current step. Since the first time through we dropped something into driver_internal_info, it says "oh I must have been booted up and the changes applied. hooray."18:12
cardoeIn the "apply pile of BIOS settings" case, we check if in fact those settings were applied and if not we go into error.18:12
cardoeIn the "reset BIOS" case we just assume that the BIOS must have been reset.18:12
cardoeLooking to provide an escape hatch for https://docs.openstack.org/ironic/latest/admin/drivers/idrac.html#pxe-reset-with-factory-reset-bios-clean-step18:13
cardoeBecause today if we use UEFI, I cannot reset the BIOS because the box doesn't come back up.18:14
cardoeIf I'm PXE/HTTP booting.18:14
cardoeNow my patch doesn't actually address that (because I didn't touch the BIOS reset case).18:14
JayFAside: WTF WHY WOULD DELL DEFAULT TO BIOS BOOT18:14
JayFoh, I see18:15
cardoeThey don't. If you're in UEFI mode then the boot device enable/disable state comes from the BIOS settings. If you're booting in BIOS mode the PXE enable/disable comes from outside of that path.18:16
JayFyeah I read the link after making the statement. very jay of me :D 18:16
cardoeRight now I'm really just interested in we roll in a rack of boxes. They don't have HTTP boot enabled (or PXE but we're using HTTP boot). So I can't ingest them into Ironic.18:17
cardoeSo I first need to set some BIOS settings then ingest them into Ironic.18:17
cardoeJulia saw James's talk where he mentioned some of this "first touch" steps.18:18
JayFbut you can't set bios settings, because of how the flow works, without having IPA booted18:18
cardoeYep18:18
JayFI still feel like this almost would be nicer composed18:18
JayFlike some agentless apply configuration step18:18
JayFthat expects *the caller* to reboot step next if that's the right thing18:19
cardoeI’m down for doing whatever is correct. In other things we follow the task.18:28
opendevreviewMerged openstack/ironic unmaintained/yoga: Fix actual size calculation for storage fallback logic  https://review.opendev.org/c/openstack/ironic/+/93128318:48
opendevreviewMerged openstack/ironic unmaintained/zed: Fix actual size calculation for storage fallback logic  https://review.opendev.org/c/openstack/ironic/+/93128218:48
JayFI'm just thinking out loud18:48
opendevreviewMerged openstack/ironic master: Enable GLOBAL_VENV in ironic grenade jobs  https://review.opendev.org/c/openstack/ironic/+/93201618:48
opendevreviewVerification of a change to openstack/ironic master failed: Gracefully handle bad request exception  https://review.opendev.org/c/openstack/ironic/+/93184918:49
opendevreviewMerged openstack/ironic master: docs: Updates redfish metrics docs with required configuration file settings  https://review.opendev.org/c/openstack/ironic/+/93255118:49
JayFlike I alluded to before in the review jam, I'm probably not the best person to be talking design for sushy :)18:49
cardoesorry. had a repairman come by18:55
cardoehttps://opendev.org/openstack/ironic/src/commit/1f740c8a90e1c9bcdd2fb7ee589a5b38d16f8e1e/ironic/drivers/modules/redfish/firmware.py#L395-L40418:57
cardoehttps://opendev.org/openstack/ironic/src/commit/1f740c8a90e1c9bcdd2fb7ee589a5b38d16f8e1e/ironic/drivers/modules/redfish/raid.py#L991-L101218:57
cardoeThat's what I was referring to as far as the task tracking.18:58
cardoeSo on the HPE and Dell gear I've got, once you reboot the machine a task is created.18:58
JayFugh so you can't even be like "do this" and it's "okay"19:00
JayFit's literally "do this" "reboot" "okay"19:00
cardoehttps://review.opendev.org/c/openstack/releases/+/933056 is that correct?20:03
iurygregorycardoe, depends are you planning a new release for Dalmatian? if yes, yup looks correct 20:03
cardoeyeah was looking to make one final release with Python 3.8 support20:04
cardoebefore we merge the removal of Python 3.8 support20:05
iurygregorywait, in this case we need Epoxy release20:05
iurygregorynot Dalmatian, I would say20:05
iurygregoryand the only diff of commits is:20:06
iurygregoryChanges between 5.2.0 and d39c71320:06
iurygregory* d39c713 2024-09-06 13:09:30 +0000 Update TOX_CONSTRAINTS_FILE for stable/2024.220:06
iurygregory* fc5292e 2024-09-06 13:09:28 +0000 Update .gitreview for stable/2024.220:06
iurygregorythe release team won't do a release because there is no code changes20:06
cardoeokay so I need to backport the stuff into stable/2024.2 first?20:07
iurygregoryif you want a new release for Dalmatian yes, we need backports =)20:07
iurygregoryaren't we just planning on dropping 3.8 in Epoxy?20:08
iurygregoryif this is the case we need to release 5.3.0 in Epoxy20:08
cardoeYeah probably20:09
opendevreviewDoug Goldstein proposed openstack/sushy stable/2024.2: drop usage of pkg_resource for newer python compat  https://review.opendev.org/c/openstack/sushy/+/93305820:10
cardoeSo like ^ that first and then re-do that 5.2.1 with the commit ID that lands.20:11
cardoesushy doesn't really seem tied to OpenStack releases to me20:11
iurygregoryok I'm puzzled, why we would need https://review.opendev.org/c/openstack/sushy/+/933058 in Dalmatian? 20:12
cardoeSo I'm gonna speak PyPi packages if that's okay and maybe you can tell me where I'm wrong?20:13
iurygregorybecause looking at the setup.cfg we support till 3.11 in Dalmatian20:13
iurygregoryso things shouldn't break20:13
iurygregory.-.20:14
iurygregoryhttps://github.com/openstack/sushy/blob/stable/2024.2/setup.cfg20:14
cardoeSo we released sushy 5.2.0 before we added the Python 3.8 -> 3.12 support.20:14
iurygregorycorrect20:14
cardoeWe said we'd drop Python 3.8 from sushy but we wanted 1 release (5.2.1) which supported both as a migration path.20:14
cardoeSo 5.2.1 would support 3.8 -> 3.1220:14
opendevreviewPavlo Shchelokovskyy proposed openstack/ironic master: Expand detected strings in check_cipher_suite_errors  https://review.opendev.org/c/openstack/ironic/+/93291520:14
opendevreviewPavlo Shchelokovskyy proposed openstack/ironic master: Update some ipmi config option help strings  https://review.opendev.org/c/openstack/ironic/+/93302320:14
cardoeAnd then 5.3.0 would support 3.9+ only20:15
iurygregoryso in Epoxy we have 3.8 -> 3.12 currently correct?20:15
cardoeyes20:15
cardoeBut we'll release Epoxy with 3.9+20:15
iurygregoryso we just need to release 5.3.0 in Epoxy first, after that we drop the 3.8 and do another release 5.4.0 (probably)20:16
iurygregoryat least this would be my understanding, we shouldn't backport https://review.opendev.org/c/openstack/sushy/+/933058 to Dalmatian I would say20:16
cardoeI'm just trying to land what was asked.20:17
cardoeI had to get stuff into global-requirements.txt to make a version that supported 3.8 -> 3.1220:17
iurygregorywhat was added to global-requirements?20:18
cardoethe pkg_resources backport20:18
iurygregoryok, found the patch20:18
JayFI am +1 to this backport, fwiw, as long as it's in global reqs for 2024.220:20
cardoeit is20:20
iurygregoryok, it's present in the requirements for Dalmatian20:20
iurygregorymy concern is basically that in Dalmatian we don't support py3.12, so no need for the https://review.opendev.org/c/openstack/sushy/+/933058 20:21
iurygregoryand for Epoxy we can follow the idea of doing a release now before dropping py3.8 (but the version will be 5.3.0 probably, not 5.2.1)20:21
cardoeI just wanted sushy to support Python 3.12. Folks insisted that I needed to land something that supported Python 3.12 AND Python 3.8.20:23
cardoetkajinam: 20:23
cardoetkajinam: has proposed a change that basically drops Python 3.8 and removes all the backport stuff I did.20:23
cardoeWhich we want to merge for Epoxy20:23
iurygregoryI understand20:23
cardoeI'm just trying to figure out what I need to do to move forward and I don't understand.20:24
iurygregorybut we can have a release in Epoxy with the 3.8 -> 3.1220:24
JayFMy suggestion is: 1) backport to 2024.2, 2) release 2024.2, 3) maybe also release master as an epoxy deliverable before tkaj's changes land20:24
iurygregoryand after having this release we can merge tkajinam patch to drop 3.8 and have a new release in Epoxy also20:24
JayFI am skeptical anyone wants to run the release #3 would be creating, but it's probably nice for completeness anyway20:25
iurygregoryJayF, i'm puzzled why we need in 2024.2 (since we don't support 3.12 there)20:25
JayFiurygregory: 'need' is not really the measuring stick I'm using for this; we have an operator who expressed a need (cardoe) to have it, and I don't see any harm whatsoever it'd cause to meet that need20:26
cardoeSo that global-requirements piece is being removed from epoxy.20:26
JayFplus I assume if he wants it, other folks would20:26
cardoeSo the only place I can make this release would be in 2024.220:26
JayFooh20:26
cardoebecause epoxy has dropped 3.820:26
JayFthat's the winner then, and only backport+release no master release in that case imo20:26
iurygregorycardoe, are you running py3.12 in Dalmatian? (just trying to understand your case)20:26
cardoeI'm running "pip install sushy"20:27
cardoeI just want a release that works on Python 3.1220:27
JayFcardoe: if you don't care that it /also works on 3.12/ theoretically a release from master after tkaj's change merges would do that20:27
opendevreviewPierre Riteau proposed openstack/tenks master: Use openstackdocstheme for release notes  https://review.opendev.org/c/openstack/tenks/+/93305920:28
JayFbut honestly I also feel bad if the code you wrote for that bridge is just ... never in a useful place20:28
cardoeyeah. you guys flogged me over it and I had to harass the requirements repo owners to land that backport piece.20:28
cardoeonly for it to be reverted out weeks after the release.20:29
JayFyou guys is not a unanimous thing in an oss community fwiw20:29
cardoeyeah fair20:29
JayFbut like I said, I'm personally +2 to backporting and releasing20:29
cardoeiurygregory: I'm pushing our folks and teams to use upstream sushy and contribute to sushy.20:29
cardoeI've got a Python 3.12 need for making that happen.20:29
JayFiurygregory: cardoe: I'm going to +2 the backport, if you feel strongly it shouldn't be backported lets have that discussion in the change. I don't wanna get in someones' way, especially when they are trying to prove out OSS contribution (and I personally have been on the business end of that business' impatience and would like to save cardoe from that)20:31
iurygregorycardoe, right, I understand, in my mind 3.12 is something for Epoxy so our release should be in Epoxy not Dalmatian20:32
cardoeSo fwiw tkajinam's patch is what I proposed at first.20:32
JayFiurygregory: the platform stuff published by the TC is a minimum20:32
iurygregoryhappy to +1, but I don't feel like I would +W20:32
JayFiurygregory: I'm very very happy to go above and beyond for a contributor when the effort needed for us to do it is a +2A on a backport20:32
cardoeBut I was told before I could land 3.12 support I needed to continue to support 3.820:33
cardoeSo I did that part.20:33
JayFas much as we ask folks to scratch their own itches we should be accomodating20:33
cardoeI'm fine just landing tkajinam's change and making a release too.20:33
cardoeBut I know I got pushed hard to make the backport style.20:33
JayFI think both is probably the best answer20:33
iurygregoryso maybe adding 3.12 in Dalmatian for sushy as supported?20:33
cardoesushy is used in a bunch more than just OpenStack releases.20:34
iurygregorynot only the backport, but also updating the setup.cfg there20:34
cardoeOpenShift has a bunch of repos that depend on it.20:34
JayFhttps://github.com/openstack/sushy/blob/master/setup.cfg#L23 that's a damned good point20:34
cardoeAnd don't pull in anything else OpenStack-y20:34
JayFwe can't backport declaration of support for 3.1220:34
iurygregorynot really a bunch, only 1 repo in OCP20:34
JayFand is this patch worthwhile without a setup.cfg change?20:34
JayFcardoe: iurygregory: my downstream utilizes sushy, and they aren't the only one I've heard20:35
iurygregoryJayF, yup that what I'm wondering20:35
iurygregoryin Dalmatian we are 3.8 -> 3.1120:35
JayFcardoe: and is this patch useful **to you/rax** without a setup.cfg change?20:35
opendevreviewDoug Goldstein proposed openstack/sushy master: advertise Python 3.12 support  https://review.opendev.org/c/openstack/sushy/+/93306020:35
cardoeThere's the setup.cfg change. 20:35
JayFyeah, but that is not backportable at all20:36
cardoeyes this is usable to me without a setup.cfg change.20:36
JayFwhich is what I'm asking, is a 2024.2 release useful to you without the setup.cfg change along with it20:36
JayFthen I'm still +2 to backporting it20:36
cardoetoday we're pulling a git commit via requirements.txt20:36
iurygregoryoh ok, if works without the setup.cfg I'm ok with it20:36
JayFsetup.cfg only impacts `pip install X` resolution20:36
cardoeJayF: nope it doesn't20:36
JayFoh, really?20:37
cardoeThat's just text on pypi.org20:37
JayF*blink*20:37
cardoeIt's not used by the resolver at all.20:37
JayF:-O20:37
* JayF goes to pip install ironic==120:37
cardoepython_requires is the line in setup.cfg you'd have to do20:37
cardoepython_requires=>3.8,<3.1220:37
JayFahhh that makes more sense20:37
JayFit wouldn't be python if there weren't 4 ways of doing the same thing that mildly overlap20:38
JayFmaybe add a   ^ packaging ecosystem 20:38
JayFsomewhere around ther 20:38
iurygregoryhttps://review.opendev.org/c/openstack/sushy/+/933058 +W20:39
JayFnice20:39
JayFand I learned a thing20:39
iurygregorywe learned =)20:39
JayFnext super hot python packaging thing should totally swap the metadata it uses for resolution20:40
JayFjust to mix it up a little more20:40
iurygregoryin my mind it wouldn't work without the setup.cfg XD20:40
JayFwhen someone writes plp in whatever the next rust is /s 20:40
iurygregorylol20:40
cardoeJayF: that's uv20:41
JayFthat's the one in rust, not in whatever the cool new rust will be20:42
JayFI think it's about time for another interpreted language now, which will knock everything down a peg. perl will turn into bash, php will turn into perl, ruby will turn into php, and python will turn into ruby20:43
cardoeoooo someone added httpx into global-requirements for epoxy21:15
JayFyeah, things are moving21:15
JayFand it's not as hard to get things in g-r as it used to be21:15
JayFliterally the only thing sending metrics in IPA today are four api methods21:28
JayFin a way that is not going to be useful 21:28
JayFI'm still going through the deprecation motions because I'm basically there, but there's no way anyone is using this21:28
JayFor maybe should say, there's no way this is providing value to anyone even if it's hooked up21:28
opendevreviewJay Faulkner proposed openstack/ironic-python-agent master: Vendor metrics library from Ironic-Lib & deprecate  https://review.opendev.org/c/openstack/ironic-python-agent/+/93306321:37
opendevreviewMerged openstack/sushy stable/2024.2: drop usage of pkg_resource for newer python compat  https://review.opendev.org/c/openstack/sushy/+/93305821:46
iurygregorycardoe, feel free to update the release patch since the backport is merged22:41
iurygregoryI can +1 the release22:41
opendevreviewDoug Goldstein proposed openstack/ironic master: allow running inspection hooks on redfish interface  https://review.opendev.org/c/openstack/ironic/+/93306623:05
cardoeDone. I think I did the right thing iurygregory 23:06
iurygregorychecking23:06
iurygregoryyup, correct =)23:08
cardoethat inspection hooks I've marked as a WIP for the PTG topic for tomorrow.23:11
cardoehttps://review.opendev.org/c/openstack/ironic/+/634373 JayF / dtantsur / TheJulia that's another example of the vendor specific-ness23:12
opendevreviewDoug Goldstein proposed openstack/ironic master: Enable WSGI module entrypoint for Ironic  https://review.opendev.org/c/openstack/ironic/+/92763523:29

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!