Tuesday, 2026-11-10

opendevreviewGoutham Pacha Ravi proposed openstack/governance master: Drop expired extra-acs, re-propose election officials  https://review.opendev.org/c/openstack/governance/+/101485306:37
opendevreviewGoutham Pacha Ravi proposed openstack/governance master: Add SIG chairs as extra-acs  https://review.opendev.org/c/openstack/governance/+/101485406:37
opendevreviewGoutham Pacha Ravi proposed openstack/governance master: Archive the First Contact SIG  https://review.opendev.org/c/openstack/governance/+/101485506:37
opendevreviewArtem Goncharov proposed openstack/governance master: Add Rust use case for a new IAM service  https://review.opendev.org/c/openstack/governance/+/99865209:51
gouthamrtc-members: our weekly irc meeting will be hosted here in ~30 minutes16:31
gouthamrthe agenda is here: https://etherpad.opendev.org/p/openstack-technical-committee-meeting16:32
cardoeI'll be running this week's meeting16:37
cardoe#startmeeting tc17:00
opendevmeetMeeting started Tue Nov 10 17:00:12 2026 UTC and is due to finish in 60 minutes.  The chair is cardoe. Information about MeetBot at http://wiki.debian.org/MeetBot.17:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.17:00
opendevmeetThe meeting name has been set to 'tc'17:00
cardoeWelcome to the weekly meeting of the OpenStack Technical Committee. A reminder that this meeting is held under the OpenInfra Code of Conduct available at https://openinfra.dev/legal/code-of-conduct.17:00
cardoeToday's meeting agenda can be found at https://etherpad.opendev.org/p/openstack-technical-committee-meeting17:00
cardoe#topic Roll Call17:00
gouthamro/ hello :) 17:00
dansmitho/17:01
cardoenoted absence: g o u t h a m r, s p o t z, m n a s i a d k a17:01
gouthamrcardoe: ty for getting us started - i was asking for help with next week though - not today's meeting .. but happy to sit in :D17:01
ChristianBerendt[m]o/17:01
cardoeWhelp. Practice run for me!17:01
frickler\o17:01
gouthamrcontext: OpenInfra Day NA is next week (Nov 17) 17:02
gouthamrall good cardoe; ty for doing it!17:02
cardoecourtesy-ping: gtema, noonedeadpunk 17:02
gtemao/17:03
noonedeadpunko/17:03
cardoe#topic Last week's action items17:03
cardoeNone have really carried over17:04
cardoe#topic New OpenInfra AI policy17:04
cardoe#link https://openinfra.org/legal/ai-policy 17:05
cardoeAnyone have any comments they would like to bring up?17:06
cardoeIt's been a recent hot topic so surprised that there's no concerns.17:07
gouthamri think there are several concerns17:07
gouthamrfrom prior comments, and some agreed-upon aspects that are not yet in that resolution17:08
fricklerworth to mention https://review.opendev.org/c/openstack/governance/+/1012080 in that context17:08
gouthamr#link https://review.opendev.org/c/openstack/governance/+/1012080 17:08
cardoeThat's the link I was looking for.17:09
gouthamr(that was supposed to be associated with the topic)17:09
gouthamrJayF: got us started with this17:09
gouthamrand i plan to take a stab at adding the missing bits17:09
cardoeOne thing I think worth mentioning is that folks using LLMs with security bugs. As soon as you run that issue through a non-local LLM. You've effectively disclosed that vulnerability.17:10
fungiyes, we've gone back and forth on that point with people17:11
dansmithTBH, I think it's going to be hard to not do that17:11
dansmithlike I agree, but there's also a lot of "fight fire with fire" talk about addressing the tidal wave with the "best tools available"17:12
fungiusing a remote llm service to compose a vulnerability report or with an mcp integrated into your ide when developing patches effectively discloses the details to a third party17:12
fungiif we're going to allow it, we need to be clear that we're okay with that risk17:12
dansmithit totally is disclosing, what I'm saying is:17:13
fungias well as the risk that some llm services may effectively disclose things they know to other users of the service (e.g. get trained on the things they're prompted by)17:13
dansmithI think that whether we're okay with it or not, and whether we bless it or not, it is and will happen17:13
cardoeWith every new technological invention, it's own pandoras box is opened.17:13
noonedeadpunkI'd say that'r aroudn 90% chance that this vulnarability was found with online LLM17:13
noonedeadpunkSo it was not private to begin with...17:14
cardoeYeah I agree with dansmith. It's going to happen. What I think is worth doing is pointing that out explicitly in our AI doc.17:14
fungiand 50% of those are "did you know that if you obtain someone's login credentials you can boot a vm in their account, up to their allowed quota?!?"17:14
dansmithI also think it's fine if we officially don't allow it for a CYA reason, as long as we're clear that it will happen anyway17:14
dansmithand I think it will be hard to track down such a disclosure to a person or tool that was used17:15
JayFfungi: dansmith: cardoe: Where I want clarity is for contract/internal LLMs17:15
JayFI can agree that public LLM + security vuln is a form of disclosure17:15
dansmithnoonedeadpunk: right, that's why I mean by the "fire with fire"... most of these are coming from an LLM somewhere anyway :)17:15
JayFbut if someone has an internal, but shared, cluster (as I understand IBM may have?), this seems like a middle ground that is more okay17:15
clarkbone thing to keep in mind is that people have gotten fairly confrontational over onerous reporting requirements or lack of feedback on reported issues. To the point of dropping rce 0 days against software17:15
clarkbso if you say "you cannot use these tools" then people are probably just as likely to 0 day you as they are to report it the way you want them to17:16
dansmithclarkb: yeah, good point17:16
cardoeclarkb: So to me it's more of a "be aware that using a third-party LLM is equivalent to discussing the vulnerability with a third-party"17:17
noonedeadpunkthat's fair note to have 17:17
fungiwe effectively can't prevent "researchers" from using whatever tools they like when reporting suspected vulnerabilities to us, any guidance we provide is going to be to our project maintainers and downstream users/distributors later in the process anyway17:18
dansmithI think there is also a nuance here:17:19
dansmithpasting the content of the private bug into a chat tool and saying "fix it" is different than giving it specific directions like "change operation Y to ensure that condition X is met before proceeding"17:20
JayF++ (oh no, not again)17:21
cardoeWell two random atoms decided to collide in the universe... aka jay and dan agreed and the convo has slowed17:22
cardoeso let's move on17:22
fricklerbut is it really? afaict we treat the proposed fix with the same confidentiality as the bug report?17:22
cardoefrickler: I waited a bit to see if more came in but will move on17:23
cardoe#topic Gate Health17:23
cardoeAnyone have a good check on that?17:25
cardoe#link https://github.com/tox-dev/tox/issues/435717:26
cardoeWas closed which was the tox regression17:26
cardoeWe've moved past that but I know a number of projects pinned17:26
cardoeThere's a lot of things running right now and not much of an uptick in issues I've seen17:27
clarkbfrom the infrastructure side of things we've got ~2-3 issues of varying annoyance impacting jobs. One is a cloud region (you may be able to help with that cardoe I pinged Dan With on Matrix but haven't heard back yet) that seems to lose networking post boot for ~2 minutes which causes jobs to retry when ansible can't gather facts. ANother is same cloud another region we may have17:27
clarkbevidence of occasional disk image corruption causing the git repo cache to be invalid for some repos occasionally. Both are weird and I think we're sort of running out of idaes for how we can debug this as users.17:27
clarkbThere is also the centos 10 stream zswap issue that got "fixed" by disabling zswap on that platform in devstack17:27
cardoeI've raised your comments to folks.17:29
cardoeAlright well moving on then.17:30
cardoe#topic TC Tracker17:30
cardoe#link https://tinyurl.com/os-tc-dash17:31
cardoeSo we've got a few things still out there.17:31
cardoe#link https://review.opendev.org/c/openstack/governance/+/99865217:31
cardoegtema's proposal has been reborn for a new IAM service17:32
cardoe#link https://review.opendev.org/c/openstack/governance/+/99276517:32
cardoeWe've been discussing tested runtimes for clients and libraries17:32
cardoeAnd there's a few where we are refreshing the SIG chairs and membership17:33
cardoe#link https://etherpad.opendev.org/p/tc-2027.1-tracker17:34
*** gmaan is now known as gmaan_afk17:34
cardoeAlong those lines with the SIGs, we're looking this cycle to refresh them and their status17:34
cardoe#link https://governance.openstack.org/tc/reference/technical-vision.html17:34
cardoeclarkb and gouthamr are spearheading an update to the technical vision document which I'm sure they'd appreciate feedback on17:35
gouthamrbarely; but, that's a good call out for me to get back to that work17:35
cardoe#link https://governance.openstack.org/tc/goals/proposed/remove-eventlet.html17:35
gouthamron the SIGs especially, we need someone to do some light dusting, and sunset SIGs that are inactive17:36
cardoeGood call out.17:36
JayFHonestly, we might want to sunset the security sig or more clearly integrate it with the VMT17:37
gouthamr#link https://review.opendev.org/c/openstack/governance/+/1014855 (Archive the First Contact SIG)17:37
JayFtwo separate stacks of security org made more sense when we had more people milling around doing security work "unofficially"17:37
gouthamrdiablo_rojo could use your eyes on the FCG change above ^ 17:37
fungiJayF: the vmt was considered part of the security sig, but yes now the vmt is essentially the whole of the active security sig participants most of the time17:38
gouthamrbut, wouldn't we want to keep it separate so we can encourage folks joining and helping with guidelines?17:38
gouthamrand any non-embargo related activities17:38
fungithose would be my questions: do we want to encourage security efforts outside the vmt's activities, and if so do we think keeping a sig around is too much effort if it doesn't actually bring them in?17:39
JayFTBH, I think those are theoretical questions. The reality is that there's been minimal engagement outside of VMT folks. I am not saying we shouldn't have an entrypoint, but if it's the same people, just call it one thing instead of two17:40
JayFI think of the work Sofia did to migrate OSSNs: that could probably have been considered technically under security sig or something, but reality is it was just a thing that needed doing, so I asked someone to do it17:40
JayFI just want less empty front doors for people to knock on and be disappointed; and that's kinda how I view inactive SIGs17:41
fungiwe also have a separate problem, wherein the vmt's authority is implicit not explicitly delegated from the tc. we made it part of a sig to sort of try to ignore that17:41
gouthamrVMT's front door is like a speakeasy17:41
fungibeing in a sig makes the vmt's activities at least an official part of openstack17:42
gouthamri kinda like that framing: VMT is a specialized part of the wider security effort17:42
JayFAgain I point at: what wider security effort? Activities not happening in VMT are generally happening at project team level17:43
JayFthe structure we have should reflect us, rather than us trying to maintain a structure for what we wish we were17:43
gouthamri was hoping that we'd do the threat modeling, writing guidelines, documenting RBAC etc through the SIG, publicly, and loosely organizing from volunteers that are interested in these various efforts 17:43
fungiin the past the security sig has engaged in goals like secure rbac and privsep, or pop-up efforts like image encryption17:43
gouthamri couldn't think of any other logical place for that sort of work17:43
gouthamrcardoe: i pinged ralonsoh again in the past week to get an update on the privsep goal17:46
cardoeIt's a lot of overlap and would the membership be different?17:46
cardoe#link https://governance.openstack.org/tc/goals/selected/migrate-to-privsep.html17:47
cardoeDid we get any updates?17:47
gouthamrnah :| i'm struggling with this17:47
gouthamri'm kinda wondering if we should change goals to require a group of people17:47
fungitie them more explicitly to the pop-up team concept?17:48
gouthamryes. it feels unreasonable to expect a single goal champion to always be available and caring/feeding for a monstrously large effort sometimes 17:48
gouthamri'll propose that, and see if you think it is reasonable17:49
cardoeThat makes sense17:50
cardoeI'll move us along.17:50
cardoe#topic Open Discussion17:50
cardoeThe only item here is to be aware of elections coming up and shorter time lines17:51
gouthamryes; i haven't had a chance to catch slaweq/ianychoi and check if they'd like to continue as election officials17:52
gouthamrChristianBerendt[m]: i proposed https://review.opendev.org/c/openstack/election/+/1014592 17:52
gouthamr^ it sets up directories for the elections for all the PTL lead teams that exist today 17:53
gouthamri think we ought to get started with the election config and put up the dates asap so we can let people plan.. 17:53
gouthamrwe can chat about this on #openstack-election if you'd like17:54
gouthamrif there are any other volunteers to help coordinate elections, please join that channel17:54
fungi#link 2027 OpenInfra Foundation Individual Member Representative nominations are open. https://lists.openinfra.org/archives/list/foundation@lists.openinfra.org/thread/3GWDHFR6VX56EZYVLE6ISIWP5ZPKI4WP/17:55
fungithose too17:55
gouthamr(i'll put out a call for election official volunteers on the ML as well)17:55
gouthamrah; sorta close in timeline17:56
gouthamr2027 Board elections will "be held Monday, January 11, 2027 at 00:01 UTC through Friday, January 15, 2027 at 23:59 UTC"17:56
gouthamrlets plan this further in #openstack-election17:58
gouthamrnext week is OpenInfra NA17:59
gouthamrcardoe has graciously agreed to run the meeting on Nov 17th18:00
gouthamris anyone here planning to be at the event?18:00
fungii think diablo_rojo is?18:00
gouthamrah neat18:00
cardoeWell we're at time as well.18:01
cardoe#endmeeting18:01
opendevmeetMeeting ended Tue Nov 10 18:01:21 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)18:01
opendevmeetMinutes:        https://meetings.opendev.org/meetings/tc/2026/tc.2026-11-10-17.00.html18:01
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/tc/2026/tc.2026-11-10-17.00.txt18:01
opendevmeetLog:            https://meetings.opendev.org/meetings/tc/2026/tc.2026-11-10-17.00.log.html18:01
cardoeThanks all.18:01
gouthamrthanks cardoe!18:01
fungis p o t z  was planning to be there as well, i thought18:02
fungiah, yeah during last week's meeting in #openinfra-diversity she confirmed as much18:03
*** gmaan_afk is now known as gmaan19:04
opendevreviewBrian Haley proposed openstack/governance master: Deprecate networking-sfc  https://review.opendev.org/c/openstack/governance/+/101496820:01

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!