Tuesday, 2022-07-05

rpittaugood morning ironic! o/07:33
opendevreviewPierre Riteau proposed openstack/tenks master: Remove Python 2 leftovers  https://review.opendev.org/c/openstack/tenks/+/84869807:35
Kirill_Hi. i'm a new dev and now working on vnc console for Ironic. i used this as a start https://review.opendev.org/c/openstack/ironic/+/396661. And have a workable version. But i also know about solution with Graphical Console Interface (https://review.opendev.org/c/openstack/ironic-specs/+/306074).i'm a bit confused who can help me with this and should we discuss it here or talk in another channel?) Thanks07:50
opendevreviewMark Goddard proposed openstack/ironic-python-agent-builder stable/yoga: Add smartmontools package to DIB image  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/84860708:06
opendevreviewMark Goddard proposed openstack/ironic-python-agent-builder stable/xena: Add smartmontools package to DIB image  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/84860808:06
opendevreviewMark Goddard proposed openstack/ironic-python-agent-builder stable/wallaby: Add smartmontools package to DIB image  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/84860908:06
opendevreviewMark Goddard proposed openstack/ironic-python-agent-builder stable/wallaby: Add smartmontools package to DIB image  https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/84860908:07
opendevreviewSwapnil Machikar proposed openstack/ironic-tempest-plugin master: Add iDRAC Redfish sync boot mode test  https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/84870909:13
opendevreviewPierre Riteau proposed openstack/tenks master: Use yoga upper constraints for openstack-tox-py36  https://review.opendev.org/c/openstack/tenks/+/84873109:32
arne_wiebalckGood morning rpittau and Ironic!09:34
rpittauhey arne_wiebalck :)09:34
arne_wiebalckLocation-aware scheduling, e.g. making sure instances go to a specific rack in the DC, what do people do to achieve this?09:35
arne_wiebalck(the best I have come up with so far is to have per rack resource classes and flavors ... is there a better way?)09:36
opendevreviewPierre Riteau proposed openstack/tenks master: Use yoga upper constraints for openstack-tox-py36  https://review.opendev.org/c/openstack/tenks/+/84873110:32
dtantsurKirill_: hi and welcome!10:55
dtantsurKirill_: I think the graphical console work is the right foundation to build on. The iDRAC proof-of-concepts relied on old-style drivers that don't even exist any more.10:55
dtantsurSince serial and graphical consoles are not mutual exclusive, we need to have both at the same time.10:56
dtantsur(and yes, this is the right channel to discuss it)10:56
Kirill_got it. So some  my thoughts: we can create a new graphical console for vnc. But we also have rdp console(for example) and it also graphical. should we relocate it to new interface?11:08
Kirill_whould not it be better to stay in one interface (console interface) and give a possibility to activate several consoles. which are not mutual exclusive. we already providing console type during get/enable requests11:10
Kirill_@dtantsur (mark check)11:17
opendevreviewJacob Anders proposed openstack/ironic master: Prevent clear_job_queue and reset_idrac failures on older iDRACs  https://review.opendev.org/c/openstack/ironic/+/84685911:27
jandersajya dtantsur hopefully ^ should be all good to go11:28
jandersthank you for your reviews 11:28
dtantsurKirill_: an ability to have several console interface implementations active at once will be much more work than finishing the (already approved by the community) idea of a graphical console interface.11:30
jandersarne_wiebalck I'm trying to track down an eTag related issue on a Lenovo server. It seems it's somehow being given two eTags (weak and strong) and not liking it: https://bugzilla.redhat.com/show_bug.cgi?id=2084059#c1511:39
jandersdo you have any thoughts on this? haven't seen this failure mode before11:39
janders'If-Match': 'W/"553b2ac3d03428a98ef","553b2ac3d03428a98ef"' - seems it is being fed same eTag as weak and strong for some reason11:40
iurygregorygood morning Ironic11:41
jandershey iurygregory o/11:41
Kirill_thanks dtantsur(sorry i can't get how to answer here). Got it11:54
opendevreviewDmitry Tantsur proposed openstack/bifrost master: [WIP] Stop passing ironic_url to ansible  https://review.opendev.org/c/openstack/bifrost/+/84874112:24
opendevreviewMerged openstack/ironic-lib master: Remove python version 2 pkgs from bindep  https://review.opendev.org/c/openstack/ironic-lib/+/83031912:26
opendevreviewJacob Anders proposed openstack/ironic master: Prevent clear_job_queue and reset_idrac failures on older iDRACs  https://review.opendev.org/c/openstack/ironic/+/84685912:27
jandersthank you for the review ajya12:34
jandersdtantsur iurygregory when you have time, can I please ask for some +2s for ^?12:34
ajyawelcome, janders 12:34
iurygregoryjanders, sure will look today12:36
jandersiurygregory ty!12:37
jandersarne_wiebalck further to the eTag question above, myself and dtantsur are wondering if we may need to consider splitting the strong and weak eTags and try them separately...12:39
iurygregoryI think we had a scenario where we need both ... (this seems like a deja vu...)12:42
jandersiurygregory one vendor's fix is another's vendor breakage... hard to win this eh?12:43
iurygregoryyes!12:43
janderswe may try all combinations...12:44
jandersbrute force approach12:44
dtantsurI *hope* we're fine trying them separately.. assuming they don't somehow change after each attempt....12:44
jandersuh oh12:44
jandersI think that may be a possibility12:44
jandersonce I figure that out I think I won't need to carry house keys anymore12:44
janderswill just pick the locks12:44
janders:D12:44
jandersthere was that movie about a guy cracking some legendary safes, will find the name12:45
janderswe're not quite at that level yet but feels like turning in that direction12:45
mraineriarne_wiebalck: I can look into pushing it to EPEL; haven't done that before so I may need some help12:46
mraineriBut I can certainly get it in GitHub as a starting point12:46
jandershttps://www.imdb.com/title/tt13024674/ < that's the movie I had in mind12:47
iurygregoryjanders, don't get yourself locked out of your house =P12:48
jandersiurygregory will try not to12:49
janderssee you tomorrow Ironic o/12:50
rpittaubye janders o/12:50
jandersthank you for the +2 rpittau12:50
rpittaunp :)12:54
opendevreviewMartin André proposed openstack/ironic stable/ussuri: Fix installation of virtualbmc with py36 in devstack  https://review.opendev.org/c/openstack/ironic/+/84754812:57
derekh__TheJulia: I missed your ping a few days ago, you were asking about backporting of https://review.opendev.org/c/openstack/ironic-python-agent/+/84617713:22
mraineriarne_wiebalck: Not having success with getting the RPM files pushed to PyPI (they get rejected as an invalid distribution)13:23
mraineriBut I have them posted on the release of the project on GitHub: https://github.com/DMTF/Redfish-Tacklebox/releases13:23
mraineriSo, that's all set for you to try out13:23
dtantsurI'm quite sure RPMs are not distributed via PyPI..13:23
TheJuliadtantsur++13:23
dtantsurgood morning TheJulia 13:23
derekh__TheJulia: yes, ideally I think I'd like it in bugfix/8.6 and 8.3 , do I need to first propose to stable ? or just the bugfix branches? 13:23
mraineriYeah, I didn't look into it too much yet and had a suspicion that might be the case13:24
TheJuliaderekh__: tl;dr in order based upon time13:24
TheJuliaderekh__: truthfully I wouldn't mind backporting all the way to wallaby since we're seeing some networkmanager differences on centos9 which is somewhat to be expected as well13:25
* TheJulia feels like a zombie13:26
TheJuliaWe had people shooting off fireworks until after midnight13:27
derekh__TheJulia: sounds like a long night13:29
dtantsurugh13:29
ajyaHi, can these 2 backports get 2nd core reviewer - https://review.opendev.org/c/openstack/ironic/+/848450 https://review.opendev.org/c/openstack/ironic/+/848451 (thanks rpittau as first).13:29
* TheJulia ponders more coffee before first meeting13:30
dtantsurmm, coffee13:33
opendevreviewDerek Higgins proposed openstack/ironic-python-agent stable/yoga: Gather details about bond interfaces if present  https://review.opendev.org/c/openstack/ironic-python-agent/+/84874213:33
dtantsurajya: will put on my list, if nobody beats me to it13:33
ajyathanks, dtantsur 13:35
iurygregoryajya, can you push a backport of "Fix redfish-virtual-media for newer iDRACs" to ironic bugfix/19.0 and bugfix/18.1? 13:39
ajyaiurygregory: ok13:40
iurygregorytks!13:40
rpittaulooking for approval for the ansible bump in bifrost if anyone has a minute https://review.opendev.org/c/openstack/bifrost/+/843847 thanks!13:41
iurygregoryrpittau, looking now13:42
opendevreviewMerged openstack/ironic-lib master: CI: Removing job queue  https://review.opendev.org/c/openstack/ironic-lib/+/84261213:52
opendevreviewAija Jauntēva proposed openstack/ironic bugfix/19.0: Fix redfish-virtual-media for newer iDRACs  https://review.opendev.org/c/openstack/ironic/+/84871714:09
opendevreviewAija Jauntēva proposed openstack/ironic bugfix/18.1: Fix redfish-virtual-media for newer iDRACs  https://review.opendev.org/c/openstack/ironic/+/84874714:10
dtantsurTheJulia: I've just realized that Columbus is in Ohio (yeah, I'm bad at the US geography), and that feels weird given the recent news..14:12
opendevreviewMerged openstack/bifrost master: Use Ansible 5.x  https://review.opendev.org/c/openstack/bifrost/+/84384714:32
TheJuliadtantsur: yeah :(14:37
TheJuliaIn more than one way, for me14:37
TheJuliadtantsur: on a plus side, Columbus is a bit of a microcosm, but still :\15:00
opendevreviewMerged openstack/ironic bugfix/20.2: Fix redfish-virtual-media for newer iDRACs  https://review.opendev.org/c/openstack/ironic/+/84859515:05
opendevreviewMerged openstack/ironic stable/xena: Fix redfish-virtual-media for newer iDRACs  https://review.opendev.org/c/openstack/ironic/+/84845015:07
opendevreviewMerged openstack/ironic stable/wallaby: Fix redfish-virtual-media for newer iDRACs  https://review.opendev.org/c/openstack/ironic/+/84845115:09
Kirill_dtantsur: one more topic to discuss) As you know vnc requires a vnc hand shake. in this case novnc_proxy needs a vnc password which is on Ironic side. In my solution: when Nova sends get_console request in answer it gets  url and also this password. on next step Nova puts vnc password into DB->'console_auth_tokens' table. then novnc_proxy takes from this table password and use it for hand shake. Is it a good solution?  15:33
Kirill_<dtantsur>: or we have another way for sharing such creds between components? thanks15:35
dtantsurKirill_: I cannot really comment for the Nova side (do they already have this table?). I don't feel very easy to publish a password in our API.. I need to remember what we do for console. TheJulia do you still have the console stuff in the context?15:36
TheJuliadtantsur: loading15:39
TheJuliathe tapes are spinning, please wait15:39
TheJuliawhiiiirrrllll beep boop.15:40
Kirill_dtantsur: about 'console_auth_tokens' table. it contains a significant infor for novnc_proxy like:token_hash,console_type,updated_at and so on. each time when user tries to access console, looks inside it. per my understanding its not a bad place for password))) but better if someone confirm it15:43
TheJuliaSo I remember rescue originally had an idea of self generation and providing the password on the api surface response. I remember we didn't really  like like and kind of swapped things around so it is always supplied. However in this case we can't really avoid that15:44
TheJuliaWe presently store the url on the request, but.. we would need a synchronous call to get console data most likely since I suspect it can't even be cached really15:45
TheJuliathey have short lifetimes generally from what I'm aware15:45
TheJuliaand I'm not really a fan of a password on a url, but I have seen some patches out there that do just that...15:47
TheJuliahmm, what do we return on get console15:49
Kirill_TheJulia: ironic provides url and console type, i've added password field.15:51
Kirill_this field is used by nova, user does not see it15:52
TheJuliainside the console info field, so nested one level down15:52
TheJuliayeah, I see it in the shellinabox driver, and a driver for VNC would be sensible as long as it is not stored. I believe they are single use... but I could be very wrong on that15:53
TheJuliaKirill_: so yeah, I suspect that is likely okay in the grand scheme of the universe if it can be returned, and is not stored...15:54
TheJulia\o/ https://github.com/openstack/ironic/blob/master/ironic/common/policy.py#L812-L82115:56
TheJuliaI planned ahead!15:56
TheJuliaeh, lessee member might be needed at some point...15:57
TheJuliaDepends on how nova, in the future once RBAC is all sorted, how it issues the request.15:57
Kirill_TheJulia: seems that i didn't get this( what do you mean "if it can be returned, and is not stored...". so the idea of storing password in 'console_auth_tokens' is not good? the problem is that  nova_compute_ironic asks url but novnc_proxy only use this table. whould you like to send password/console_info request from proxy to ironic?16:04
opendevreviewDmitry Tantsur proposed openstack/bifrost master: Stop passing ironic_url to ansible modules  https://review.opendev.org/c/openstack/bifrost/+/84874116:20
TheJuliaKirill_: storing it in ironic is not a good idea16:23
TheJuliaI have no issue nor real concern what nova does with it16:23
TheJuliaKirill_: nova makes a get console call if memory serves, so just having it in the reply as you suggested, sounds good to me16:24
arne_wiebalckjanders: that is totally possible, my fix for this was mostly to solve my problem :)16:25
arne_wiebalckmraineri: I have not done that either ... let me know if you need any help and I can get you in touch with some folks here who have done it16:25
dtantsurWill anyone be terribly upset if I deprecate the redeploy playbook from bifrost? It's not tested in the CI, and is honestly quite trivial.16:30
TheJuliano objection16:30
TheJuliaI only had it there as an example16:30
TheJulia(story of our lives and whatnot)16:30
rpittaugood night! o/16:31
Kirill_TheJulia: thanks! need to ponder what we just discussed)16:37
opendevreviewDmitry Tantsur proposed openstack/bifrost master: [WIP] Deprecate os_ironic_node_info  https://review.opendev.org/c/openstack/bifrost/+/84876516:37
TheJuliaI guess I need to get a file out, and get back to addressing the sharp edges on the anaconda driver16:53
TheJuliaHmm.. maybe I need a large grinder16:56
dtantsursee you tomorrow folks o/17:15
TheJuliao/17:17
TheJuliaJayF: so I was looking at https://review.opendev.org/c/openstack/ironic/+/834709/5/ironic/common/images.py, and wondering if we should just issue a HEAD request17:34
TheJuliathat should tell us *for sure*17:34
TheJuliaand technically the glance call already does the same afaik17:34
TheJuliaso it is not like it would be a huge behavior change...17:35
* TheJulia ponders17:49
opendevreviewMerged openstack/ironic-python-agent master: CI: Removing ironic job queue  https://review.opendev.org/c/openstack/ironic-python-agent/+/84260518:10
TheJuliadtantsur: I seem to remember recently chatting about nuking the adoption requirement for rebuild to be functional... am I remembering correctly21:57
opendevreviewJulia Kreger proposed openstack/ironic master: Make anaconda non-image deploys sane  https://review.opendev.org/c/openstack/ironic/+/83470922:04

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