15:00:13 <TheJulia> #startmeeting ironic
15:00:14 <TheJulia> o/
15:00:14 <openstack> Meeting started Mon Sep  7 15:00:13 2020 UTC and is due to finish in 60 minutes.  The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:15 <iurygregory> o/
15:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:17 <TheJulia> Good morning everyone!
15:00:19 <openstack> The meeting name has been set to 'ironic'
15:00:23 <dtantsur> o/
15:00:27 <kaifeng> o/
15:00:28 <martalais> o/
15:00:32 <TheJulia> Our agenda can be found on the wiki as always!
15:00:33 <TheJulia> #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting
15:00:34 <arne_wiebalck> o/
15:00:35 <ajya> o/
15:00:57 <hjensas> o/
15:00:57 <TheJulia> #topic Announcements / Reminders
15:01:40 <TheJulia> #info CFP is open for the Open Infra Summit's Forum component (The part where we discuss larger problems and possible solutions). If anyone has ideas, please feel free to submit sessions via the session submission tool.
15:01:48 <TheJulia> #link https://cfp.openstack.org/
15:02:16 <TheJulia> #info Doodle regarding kexec discussion is available, however we've passed ~half the options at this point
15:02:23 <TheJulia> #link https://doodle.com/poll/ihbbtns36b2rkhfd
15:02:35 <TheJulia> I'll try and close that out early tomorrow and get something scheduled
15:03:05 <TheJulia> #info Ironic has been submitted to Boston University's open source mentorship program.
15:03:28 <TheJulia> Specifically the proposed focus area is kexec + long lived agents, but is generally painted as needs of the scientific and research communities using ironic.
15:03:36 <TheJulia> We should have news on that this week actually
15:03:37 <openstackgerrit> vinay kumar muddu proposed openstack/ironic master: Allow HttpImageService to accept custom certificate  https://review.opendev.org/742936
15:03:43 <iurygregory> awesome!
15:03:54 <TheJulia> #info Reminder: Please self triage bugs!
15:04:03 <TheJulia> #link https://docs.openstack.org/ironic/latest/contributor/bugs.html#triaging-guide
15:04:11 <TheJulia> I think that is it for announcements/reminders
15:04:14 <dtantsur> this ^^ especially concerns cores and long-term contributors
15:04:15 <TheJulia> Does anyone have anything else?
15:04:16 <dtantsur> well
15:04:20 <iurygregory> one more =)
15:04:29 <dtantsur> we have deadlines of ironicclient and ironic-inspector-client releases this week, I think
15:04:51 <iurygregory> we have a new contributor that will be adding the support for volume target on openstack sdk - martalais =)
15:05:18 <dtantsur> once more: welcome martalais :)
15:05:24 <arne_wiebalck> Also: doodle for regular bare metal SIG slot https://doodle.com/poll/3hpypw73455t2g24
15:06:10 <TheJulia> Looks like the client libraries are good for release (one in process of merging for python-ironicclient
15:06:23 <martalais> thank you! =)
15:06:29 <TheJulia> #link https://doodle.com/poll/3hpypw73455t2g24
15:06:59 <arne_wiebalck> thanks TheJulia
15:07:14 <TheJulia> Also, it is a holiday in North America today, so I expect we can't really reach full consensus today, but we can kind of bring awareness and so on and so forth.
15:07:18 <openstackgerrit> Merged openstack/ironic master: Remove token-less agent support  https://review.opendev.org/749407
15:07:23 <openstackgerrit> Merged openstack/ironic-inspector master: Identify accelerator devices during introspection  https://review.opendev.org/745289
15:08:07 <TheJulia> No action items from the prior week so we can skip ahead to subteam status reports
15:08:23 <TheJulia> #topic Review subteam status reports
15:08:26 <TheJulia> #link https://etherpad.openstack.org/p/IronicWhiteBoard
15:08:58 <TheJulia> Starting at line 300!
15:09:17 <TheJulia> As victoria is quickly coming to a close!
15:09:53 <TheJulia> iurygregory: what i the latest on zuulv3/grenade. I know we dsicussed it last week
15:10:08 <iurygregory> TheJulia, the latest is the multinode grenade
15:10:21 <iurygregory> going to update the whiteboard
15:10:42 <TheJulia> k
15:11:36 <TheJulia> dtantsur: any luck with keystone folks regarding: https://review.opendev.org/#/c/747670/ ?
15:11:36 <patchbot> patch 747670 - keystoneauth - Add a missing get_auth_ref implementation for noop... - 1 patch set
15:11:40 <openstackgerrit> Merged openstack/sushy master: Update master for stable/victoria  https://review.opendev.org/750159
15:11:41 <openstackgerrit> Merged openstack/sushy stable/victoria: Update .gitreview for stable/victoria  https://review.opendev.org/750157
15:11:41 <openstackgerrit> Merged openstack/sushy stable/victoria: Update TOX_CONSTRAINTS_FILE for stable/victoria  https://review.opendev.org/750158
15:11:47 <dtantsur> TheJulia: did not get to pinging them
15:12:08 <TheJulia> ok
15:12:14 <TheJulia> is this a blocker for us?
15:13:06 <iurygregory> if people are using http_basic and openstackclient yes
15:13:07 <iurygregory> =)
15:13:22 <iurygregory> ironic/inspector client works fine
15:13:49 <openstackgerrit> Dmitry Tantsur proposed openstack/ironic master: Add an option to require TLS for agent callback_url  https://review.opendev.org/749916
15:13:54 <openstackgerrit> Merged openstack/bifrost master: Use packaged iPXE ROMs for openSUSE  https://review.opendev.org/747894
15:14:08 <TheJulia> okay, so likely not a blocker, just a general thing that should eventually merge from the way I'm groking this
15:14:21 <TheJulia> I see dmitry has been super busy on ramdisk TLS \o/
15:15:14 <dtantsur> I have, yes :)
15:15:25 <dtantsur> it works, just needs rebasing now because of the agent token patch
15:15:32 <TheJulia> Regarding making CI manageable again. It feels like we have made some good progress there, It seems job renames/de-dupes are still kind of pending
15:15:41 <TheJulia> dtantsur: awesome!
15:15:56 <TheJulia> I'll be reviewing that as my #1 review priority this coming week!
15:16:30 <TheJulia> iurygregory: is the ruling that things just kind of work with dhcp-less deployments?
15:16:51 <iurygregory> TheJulia, yes!
15:16:57 <TheJulia> For privsep, I think we have to punt that to the next cycle
15:16:59 <iurygregory> dtantsur, gave the idea to add a job
15:17:23 <iurygregory> I'm going to look at how we can build and image during CI run that would allow us to run dhcp-less
15:17:32 <iurygregory> since we need the simple-init element present
15:17:39 <TheJulia> iurygregory: if you wouldn't mind also adding a note to the feedback on testing to the section
15:17:45 <TheJulia> ack
15:17:48 <TheJulia> ok
15:17:52 <TheJulia> may be easiest with bifrost, fwiw
15:18:01 <iurygregory> yeah for privsep we talked 2 weeks ago I think ...
15:18:09 <iurygregory> TheJulia, ack
15:19:04 <TheJulia> dtantsur: in-band done for the cycle?
15:19:14 <dtantsur> deploy steps? yes, I think so
15:19:37 <TheJulia> Awesome, definitely something to mention for the cycle highlights
15:20:01 <dtantsur> yup
15:20:34 <openstackgerrit> Merged openstack/bifrost master: Follow up on "Fix error 601"  https://review.opendev.org/749499
15:20:52 <TheJulia> Well, we could use a redfish interop profile update, but Richard has been heading that up and is away today due to the holiday
15:20:57 <TheJulia> I guess we're good to move on?
15:21:25 <dtantsur> iurygregory: are you going to work on vmedia support in bifrost?
15:21:37 <dtantsur> just clarifying because it's on my todo list as well, but not for right now
15:21:51 <TheJulia> Hmm, that could be a very good addition
15:21:55 <iurygregory> dtantsur it fits for the dhcp-less so I can do that =)
15:22:04 <dtantsur> yeah, that's why I'm asking
15:22:11 <dtantsur> you'll need it if you want a bifrost CI job with dhcp-less
15:22:12 <TheJulia> Good to proceed to priorities for the week?
15:22:15 <dtantsur> yep
15:22:34 <iurygregory> we are good to proceed I would say
15:22:36 <TheJulia> #topic Deciding on priorities for the coming week
15:22:48 <TheJulia> #link https://etherpad.opendev.org/p/IronicWhiteBoard
15:22:57 <TheJulia> Line 140.
15:23:01 <TheJulia> Guess we need to clean that up first
15:23:08 * TheJulia removes struck through lines
15:23:19 <dtantsur> yeah, a lot of things have merged
15:24:22 <TheJulia> Indeed, and lots of items to add :)
15:25:18 <TheJulia> so, I don't see much else and everything proposed to add at line 215 LGTM toa dd
15:25:30 <TheJulia> I didn't go through ironic too closely since we have TONS of patches that are presently failing CI :\
15:26:17 <TheJulia> Any objections? Any additional items to add?
15:26:33 <iurygregory> no objections from me
15:26:41 <dtantsur> none from me (as expected)
15:27:07 <openstackgerrit> Dmitry Tantsur proposed openstack/ironic master: Accept and use a TLS certificate from the agent  https://review.opendev.org/747136
15:28:37 <TheJulia> That looks good to m e
15:28:40 <TheJulia> to me
15:28:54 <TheJulia> Onward I guess!
15:29:18 <openstackgerrit> Dmitry Tantsur proposed openstack/ironic master: Accept and use a TLS certificate from the agent  https://review.opendev.org/747136
15:29:27 <TheJulia> #topic Discussion
15:29:32 <openstackgerrit> Dmitry Tantsur proposed openstack/ironic-python-agent master: Generate a TLS certificate and send it to ironic  https://review.opendev.org/749930
15:29:38 <TheJulia> First item up for discussion! vPTG time slots
15:29:40 <TheJulia> #link https://ethercalc.openstack.org/7xp2pcbh1ncb
15:30:41 <TheJulia> I was thinking maybe trying to run a time slot from 2-4 PM UTC on say two or three days?
15:30:50 <iurygregory> sounds like a plan
15:30:55 <TheJulia> Any thoughts? feelings? concerns?
15:31:17 <dtantsur> sounds about right. I guess 3 days works better
15:31:29 <TheJulia> We also just had our midcycle last week and didn't have TOO much come out of that which would be larger discussions
15:31:30 <dtantsur> I think we had 3 days the last time, right?
15:31:37 <TheJulia> I think we did 4
15:31:57 <dtantsur> maybe. but you're right, with regular midcycles we need less PTG time
15:32:13 <TheJulia> Yup
15:33:03 <TheJulia> Anyway, Next topic!
15:33:07 <TheJulia> raid_config behavior contract - Making the behavior non-destructive changes only!
15:33:23 <openstackgerrit> Dmitry Tantsur proposed openstack/bifrost master: Remove unnecessary overrides of ansible_python_interpreter  https://review.opendev.org/749262
15:33:46 <dtantsur> I'm worried how it's going to work when the RAID config is not specified precisely
15:33:48 <TheJulia> The discussion of dell's configuration mold feature was brought up again this past week after the midcycle.
15:33:58 <TheJulia> #link http://eavesdrop.openstack.org/irclogs/%23openstack-ironic/%23openstack-ironic.2020-09-02.log.html#t2020-09-02T22:27:47
15:34:00 <dtantsur> I guess we don't have most of Dell folks here now?
15:34:06 <ajya> only me
15:34:10 <dtantsur> right
15:34:15 <TheJulia> And tl;dr they wish that the behavior contract is changed.
15:34:33 <TheJulia> I too have the same concern, so I think I really wanted to mention this item so we're aware and we're prepared to try and discuss next week?
15:34:56 <dtantsur> or on the PTG?
15:35:08 <arne_wiebalck> For my understanding, I guess this would apply to software RAID?
15:35:11 <TheJulia> That will likely be better because there is a lot of emotion involved
15:35:16 <dtantsur> since it's pretty late to change behavior this cycle (we would have feature freeze if we were a "normal" project)
15:35:25 <TheJulia> ++
15:35:29 <dtantsur> arne_wiebalck: any RAID, although I'm not sure how
15:35:29 <ajya> PTG - end of October?
15:35:41 <arne_wiebalck> dtantsur: we were discussing this at the time I think
15:35:59 <arne_wiebalck> dtantsur: and then decided to not do this
15:36:05 <arne_wiebalck> dtantsur: back then :)
15:36:10 <dtantsur> there at least has to be a way to wipe all RAID configuration
15:36:18 <dtantsur> yep, and I think for good reasons
15:36:19 <arne_wiebalck> dtantsur: exactly
15:36:33 <TheJulia> dtantsur: this too, it is presently a read only field in the API if memory serves
15:36:46 <dtantsur> I mean, the actual RAID configuration applied to a machine
15:37:02 <TheJulia> oh, well that too
15:37:05 <dtantsur> because otherwise you can get stuck with something you cannot even delete without trying to re-create it as target_raid_config
15:37:23 <arne_wiebalck> dtantsur: ++
15:37:26 <TheJulia> yeah
15:37:43 <dtantsur> what ironic now does is a "cloud" behavior
15:37:52 <dtantsur> "I own this machine, and I own what RAID means for it"
15:38:44 <arne_wiebalck> I guess we could always make this configurable, no?
15:38:46 <dtantsur> so, this is our first (?) topic for the PTG
15:38:47 <TheJulia> I'm really worried about the edge cases this will create in trying to reconcile it
15:39:08 <dtantsur> for now I'm fine with adding such an option, I don't feel easy about making it the default
15:39:12 <dtantsur> exactly
15:39:27 <TheJulia> I don't get the motivation for forcing field reconcilliation
15:39:29 <arne_wiebalck> No, not the default, the exception :)
15:39:47 <TheJulia> bios will re-cache and we can set it to update the settings easily
15:39:59 <TheJulia> raid... is just a whole "ball of wax"
15:40:05 <dtantsur> you mean, changing raid_config based on something that is not our RAID interface?
15:40:11 <dtantsur> I'm pretty much -2 on that
15:40:20 <TheJulia> and yup
15:40:22 <dtantsur> at least based on what I've heard so far
15:40:33 <TheJulia> So PTG topic it does seem
15:41:09 <ajya> thanks for input, will discuss internally
15:41:23 <ajya> there might be more questions before PTG
15:41:32 <openstackgerrit> Merged openstack/ironic master: Switch Ironic to openstacksdk for Neutron  https://review.opendev.org/734873
15:41:35 <openstackgerrit> Merged openstack/ironic bugfix/15.2: Handle an older agent with agent_token  https://review.opendev.org/749563
15:41:41 <openstackgerrit> Merged openstack/ironic stable/queens: Fix for failure in cleaning  https://review.opendev.org/747288
15:41:46 <openstackgerrit> Merged openstack/python-ironicclient master: Remove Babel requirement  https://review.opendev.org/747581
15:42:11 <TheJulia> ajya: one thing that would be good to understand is why there is such a desire to have the field reconciliation since that is the huge sticking point
15:42:21 <openstackgerrit> Radosław Piliszek proposed openstack/bifrost master: [DNM] Dummy  https://review.opendev.org/750195
15:42:25 <dtantsur> yep
15:42:28 <TheJulia> Specifically for fields that cannot be scheduled upon
15:42:34 <ajya> TheJulia: to behave the same way as existing RAID interfaces
15:42:41 <dtantsur> but why?
15:42:56 <dtantsur> to behave the same way as existing RAID interfaces it has to be implemented by calling the RAID interface
15:43:01 <dtantsur> but rpioso doesn't seem to want that
15:43:20 <ajya> not behave, but to have the same output
15:43:26 <TheJulia> but why?
15:43:36 <dtantsur> well, as long as ironic does not care or try to interpret the output - fine
15:43:38 <TheJulia> It all comes back to how does this help the operator
15:43:51 <TheJulia> it _really_ doesn't for us to try and guess
15:43:57 <TheJulia> and reconcile configuations
15:44:18 <TheJulia> Anyway, we should move on to Baremetal SIG and RFE review
15:44:25 <dtantsur> ++ for moving on
15:44:37 <ajya> ok, will address this later
15:44:37 <TheJulia> #topic Baremetal SIG
15:44:42 <arne_wiebalck> Nothing apart from the doodle mentioned at the beginning.
15:44:44 <TheJulia> arne_wiebalck: o/
15:44:47 <TheJulia> Awesome
15:44:50 <TheJulia> #topic RFE Review
15:45:01 <TheJulia> We have one item for review, to make node.automated clean work in both directions
15:45:09 <TheJulia> #link https://storyboard.openstack.org/#!/story/2008113
15:45:56 <dtantsur> I'm not sure we should review it without JayF, he used to have a strong opinion on it
15:46:01 <TheJulia> Yeah
15:46:03 <dtantsur> but I'm ready to hear other's feedback
15:46:08 <dtantsur> others'
15:46:11 <TheJulia> I too have a strong opinion for cleaning
15:46:26 <dtantsur> so do I, but I think we ended up sending the wrong message
15:46:27 <TheJulia> but we did kind of create the situation
15:46:41 <dtantsur> if we only allow to enable cleaning per node, people will be inclined to disable it globally
15:46:53 <dtantsur> maybe we need an option or even an API policy?
15:46:53 <TheJulia> Not only message, but some cases cleaning is just not really needed in non-cloudy cases
15:47:10 <dtantsur> so that JayF could disable this policy for all nodes and prevent disabling cleaning per node?
15:47:12 <TheJulia> dunno, I do like the logging note to record it when changed
15:47:20 <TheJulia> possibly
15:47:25 <dtantsur> I'll add to the RFE
15:47:27 <TheJulia> it wouldn't make the logic terribly complex
15:47:38 <dtantsur> not at all
15:47:43 <TheJulia> I suspect we hold on to this one for next week
15:49:07 <TheJulia> Well in that case, Time for Open Discussion!
15:49:10 <TheJulia> #topic Open Discussion
15:50:26 <dtantsur> I'm going to work on the iscsi deploy deprecation
15:50:34 <dtantsur> last minute objections or comments are accepted on the ML
15:50:44 <iurygregory> go for it!
15:50:45 <TheJulia> dtantsur: sounds good
15:50:47 <TheJulia> \o/
15:51:40 <TheJulia> Well,  guess there is not much left to discuss if I'm thinking of going and finding videos of crickets :)
15:51:57 <dtantsur> ++
15:52:10 <TheJulia> Thanks everyone! Have a wonderful day and week!
15:52:31 <dtantsur> you too!
15:52:31 <arne_wiebalck> Thanks TheJulia !
15:52:41 <TheJulia> #endmeeting