Thursday, 2026-09-03

*** zigo_ is now known as zigo14:07
gouthamrhttps://bugs.launchpad.net/glance/+bug/2158998 is now public14:27
gouthamrhttps://bugs.launchpad.net/glance/+bug/2161330 is now public14:27
gouthamrhttps://bugs.launchpad.net/glance/+bug/2158999 is now public14:27
gouthamrhttps://bugs.launchpad.net/glance/+bug/2160020 is now public14:27
fricklerthx gouthamr. is there already a draft for an advisory? or will it be multiple ones?14:33
gouthamrfrickler: one advisory, working on it.. 14:33
gouthamr(https://bugs.launchpad.net/glance/+bug/2160020 will get a ride on that advisory)14:34
fungifrickler: we typically switch the bugs to public immediately before fixes get pushed to gerrit, then incorporate the urls for those changes into the advisory data, so there's a 30-60 minute lag between those steps14:41
fricklerusually (or at least sometimes?) there is a draft prepared in the bugs, which I couldn't find this time, so I was just wondering whether I missed it14:44
fungisome developers have in the past added a draft advisory in a private bug comment, the vmt process doesn't expect that, our standard is to include a draft impact description which covers the relevant details14:46
JayFThe draft OSSA in the bug is a classic Jay move for prepping for go-live :)14:47
JayFI don't trust myself to keep it locally14:47
fungii haven't looked at those bugs to see if there was an impact description reviewed in them14:47
JayFnot a VMT process, just Jay making sure the internet has all the pieces when I run the day-of14:47
fungiyeah, particularly in cases where one vmt member writes the ossa patch in advance for another vmt member to push to gerrit, including an export of the commit with relevant placeholders as a bug attachment similar to other patches can help coordination. but really the important parts should be covered by the impact description anyway and the rest is fairly mechanical15:03
JayFit's as simple as "if I'm doing something important at 6/7am local time... I need to have done all the thinking beforehand" :D 15:05
fungimakes sense. i usually just leave it as a local commit and then amend it the next morning once i have the remaining details, but i can understand not wanting to have to redo it in a pinch of something goes sideways with the machine you left it on15:07
JayFI think you manage the idea of roving workstations by using ssh to get around; I manage it by keeping as much of my state as possible in bugs/gerrit/etc (I swap pretty indiscriminately between using my desktop and my laptop)15:09
fungithat's true. i keep a tmux session going on my main workstation and then multi-attach portables to it with mosh+wireguard15:10
opendevreviewGoutham Pacha Ravi proposed openstack/ossa master: Add OSSA-2026-038  https://review.opendev.org/c/openstack/ossa/+/100383115:24
gouthamrhttps://review.opendev.org/q/hashtag:%22ossa-2026-038%22+(status:open%20OR%20status:merged)15:28
gouthamrwoah, rosmaita 15:28
gouthamrmessed up the reporter->CVE attribution, fixing that15:35
opendevreviewGoutham Pacha Ravi proposed openstack/ossa master: Add OSSA-2026-038  https://review.opendev.org/c/openstack/ossa/+/100383115:35
opendevreviewGoutham Pacha Ravi proposed openstack/ossa master: Add OSSA-2026-038  https://review.opendev.org/c/openstack/ossa/+/100383115:54
gouthamrJayF: rosmaita ^ clarified the patch ordering bit 15:55
rosmaitaOK, +W16:00
gouthamrty! working on the emails16:02
* gouthamr emails sent, self-approved on openstack-announce.. 16:10
opendevreviewMerged openstack/ossa master: Add OSSA-2026-038  https://review.opendev.org/c/openstack/ossa/+/100383116:16
gouthamrgmaan: o/ looks like a stack of glance security bug fixes rely on this one: https://review.opendev.org/c/openstack/tempest/+/100356016:30
gouthamrsorry to seek urgent reviews, but, any chance you have some time, or maybe someone else we can bug?16:30
gmaanyeah, I am on it. ws tetsing the tests in glance gate as it run there only16:31
gmaanI think I got the result from one of the job and glance-s3-multistore which also run this test seems to be broken16:31
gmaanbut I am good on that and will +w on tempest change16:31
gouthamryeah, i was scratching my head which specific job etc, but abhishek seems to have identified the job run16:32
gouthamrtyty gmaan!16:32
gmaantwo jobs run this test and one pasing and other failing but for other issue (means did not run test itself)16:32
gouthamrack16:32
gmaan+w it, let's see how fast it merge. I will keep eyes on it16:33
gouthamr\o/16:34
fungithanks for the hard work on these!16:53
gouthamr++ MITRE has been informed to publish these three CVEs.. will likely happen in the next ~24 hours 17:01
gouthamr(version numbers were updated, and Sami Yessou was added as a co-finder for CVE-2026-71196)17:01
gouthamrfungi: wanted to pick your brain on: https://review.opendev.org/c/openstack/releases/+/1003379 17:05
gouthamrso far, i've been making patch bumps whenever I requested a security release.. it is a judgement thing, and i agree with elod that this deserved a minor version bump. 17:05
gouthamrTangentially, i was glad for his -1 because i overlooked the fact that a fix that was in 2025.1/2025.2 wasn't in 2026.1 17:05
gouthamrand gtema rectified that now17:05
gouthamrbut, what's the right thing to do here?17:05
gouthamrthe fix in question isn't backwards compatible by the letter. It introduces a config opt that has a secure default, but could break someone theoretically17:06
gouthamron upgrade17:06
gouthamr(https://review.opendev.org/c/openstack/keystone/+/997369)17:06
gouthamr(this question belongs in #openstack-release perhaps, or #openstack-dev, or #openstack-keystone because of how cross-impacting it is.. but, this is a security release proposal, so seeding the discussion here. i'll take whatever takeaways bck to the patch)17:09
fungigouthamr: i'm more of a linux-style "these are still just bugs" person, but i don't object to the interpretation that recognized security fixes deserve a minor instead of only patch-level bump (sort of a blend between classic semver and pridever approaches i guess)17:09
JayFIt's worth noting that practicality has to play here too17:11
gouthamri wasn't thinking of the CVE/OSSA/OSSN impact, but more about functionality itself. will i see a new feature on upgrade? will i break? 17:11
JayFoften we don't have room, reliably, to bump at the "right level" for stable branches17:11
JayFThis is why Ironic, as policy, cuts a major version for bugfix branches, so we always have room for stable branch releases if something serious comes up17:12
gouthamrthere's room with the minor version bump here.. 17:12
fungithere are grey areas, as i pointed out in https://review.opendev.org/c/openstack/releases/+/100361117:12
gouthamri see, so the OSSAs in question used a minor bump afaict17:14
gouthamrthis one for instance:17:15
gouthamrhttps://security.openstack.org/ossa/OSSA-2026-008.html17:15
gouthamrsays: Ironic: >=4.3.0 <26.1.6, >=27.0.0 <29.0.5, >=30.0.0 <32.0.1, >=33.0.0 <35.0.117:15
gouthamrand Ironic 32.0.1 doesn't exist, 32.1.0 does17:15
JayFthat doesn't render that statement inaccurate.17:15
JayF32.1.0 is greater than 32.0.1 and less then 33, so it's got the fix17:16
JayFThere's no = on that clause implying that specific version exists17:16
fungiright, that's the approach we've operated on since the begining. the expression is correct, the expressed minimum/maximum versions aren't necessarily assumed to exist17:17
gouthamri can live with that honestly, because that's how software is versioned/built/used :P but, will it satisfy the "security advisory mentioning missing versions" thingy that cardoe brought up? 17:17
JayFI am not compelled to adjust our documented, decade+ long process, to adjust for a vibe coded github feature that didn't consider edge cases17:18
gouthamroh crap17:18
gouthamr"By convention, the < version is the next possible patchlevel release following three-component SemVer rules, so if the current highest affected version on a given branch is 8.1.0 then the affected versions includes <8.1.1 even though the next actual version tagged might well be 8.2.0 or 9.0.0 (but since we don’t know what the next version will necessarily be at the time of publication, we choose the lowest possible version 17:18
gouthamrit could be). This convention also makes it clear that prerelease versions may not be sufficient to address the vulnerability."17:18
gouthamri should RTFM17:18
fungii think that's based on invalid assumptions, we can try to better communicate the reality of our predictive versioning approach17:18
* JayF puts that note under his "VMT videos" idea board17:19
fungibasically we found a way to announce affected versions even when there is not yet a fixed version17:20
gouthamr++17:20
JayFif we tried to force just-in-time releases17:20
JayFI think I'd die from stress on day of fix 17:20
JayFlol17:20
gouthamri'm a cat i think, i've had 38 heart attacks this year :)17:21
gouthamrperfect, so i can stick to this and be consistent, and imo the minor version bump is better for the keystone releases.. 17:21
gouthamrgtema: FYI , if you're around17:21
cardoegouthamr: yes the version numbers work.17:22
cardoeI just reached out to the GitHub dad and confirmed.17:22
gtemagouthamr - small recap for me would be helpful17:23
cardoerenovate scanner, as happy as well17:23
gtemarenovate is something that would make me happy as well17:23
gouthamrgreat! 17:26
gouthamrty cardoe 17:26
gouthamrgtema: so elodilles asked why we weren't bumping the minor version for the keystone releases on https://review.opendev.org/c/openstack/releases/+/1003379; his interpretation is correct. There's a very subjective interpretation for this; and my earlier assumptions of wanting to stick with whatever we mentioned in the OSSA were fixed up by the above brainstorm 17:27
gtemaack, however I can't hold this  in my brain anymore ;-)17:28
gouthamrhaha, +1 on the patch if you agree please; so i can find someone to merge today rather than wait the weekend17:29
gtemadone17:30
gouthamrty gtema 17:32
fungilgtm still17:38
fungii might go ahead and single-rm-approve since it's getting late in europe and i may be the only one around at this point17:39
gouthamrwow17:40
gouthamrjinx :P17:40
fungiand, as you say, better to release today than wait for next week, since we avoid friday/weekend releases17:40
gouthamr++17:40
*** mrunge_ is now known as mrunge23:16

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