| chinnisreeaddagalla | Hi everyone, I’m trying to install Bifrost on an Ubuntu 22.04 ARM64 VM (running on Apple Silicon via UTM). | 00:34 |
|---|---|---|
| chinnisreeaddagalla | I was able to complete the setup and run the Ansible playbook, but I’m getting an error related to grub-efi-amd64-signed not being available. | 00:34 |
| chinnisreeaddagalla | From my understanding, this might be due to an architecture mismatch (ARM64 vs AMD64) but I wanted to confirm if Bifrost is expected to work on ARM systems or if there’s a recommended workaround. | 00:34 |
| chinnisreeaddagalla | Any guidance would be really helpful. Thank you! | 00:34 |
| opendevreview | Lennart Jern proposed openstack/sushy master: Prevent double-wrapping timeout tuple in Connector _op https://review.opendev.org/c/openstack/sushy/+/982219 | 10:09 |
| opendevreview | Lennart Jern proposed openstack/sushy master: Prevent double-wrapping timeout tuple in Connector _op https://review.opendev.org/c/openstack/sushy/+/982219 | 11:29 |
| opendevreview | Lennart Jern proposed openstack/sushy master: Prevent double-wrapping timeout tuple in Connector _op https://review.opendev.org/c/openstack/sushy/+/982219 | 11:35 |
| opendevreview | Mahnoor Asghar proposed openstack/ironic master: Soft power off before ejecting virtual media regardless of CONF.inspector.power_off https://review.opendev.org/c/openstack/ironic/+/981942 | 12:00 |
| stephenfin | JayF: RE: your question on https://review.opendev.org/c/openstack/python-ironicclient/+/981686/4/ironicclient/osc/v1/baremetal_shard.py#40, here's how I did that for OSC | 12:05 |
| stephenfin | https://review.opendev.org/c/openstack/python-openstackclient/+/970681 | 12:06 |
| stephenfin | I suspect doing the same in ironicclient would make sense | 12:06 |
| stephenfin | Also, as an aside, we'd be totally open to you merging the ironicclient OSC plugin into OSC proper, if it was an advantage. We're already planning on doing that for manila this cycle https://review.opendev.org/q/topic:%22integrate-manilaclient%22+is:open | 12:08 |
| stephenfin | ...and neutron are already in the middle of merging their stadium project cients (TaaS, VPNaaS, etc.) into OSC | 12:09 |
| JayF | stephenfin: The reason we operate our plug-in as a separate entity is that it is separately installable to create a "baremetal" (no openstack) cli for our standalone users | 13:00 |
| JayF | Thanks for the help with the shape of the types though. I knew you'd be the expert | 13:01 |
| stephenfin | Oh, you use the osc package for the `baremetal` client? Interesting. Most other clients with OSC plugins have a separate implementation for the "legacy" clients | 13:25 |
| stephenfin | (scare quotes because I realise ironic is different due to its standalone use-cases) | 13:26 |
| TheJulia | good morning | 13:29 |
| * TheJulia stumbles in like a complete and total zombie looking for... coffeee. | 13:30 | |
| JayF | stephenfin: It's not the legacy client. It literally implements the exact same commands as the openstack client does for ironic, just without the openstack prefix. | 13:32 |
| JayF | stephenfin: and while it is annoying and irrational, a lot of standalone ironic customers get really testy if they start seeing the word openstack in places | 13:32 |
| stephenfin | sorry, yes, that was my point: it's not actully legacy in ironic's case | 13:32 |
| JayF | I was just making sure you didn't think we were keeping the ironic command alive 😂 | 13:32 |
| stephenfin | nope, I'm aware. Ditto for Glance and Cinder, fwiw | 13:33 |
| JayF | I might have finally purged the last of that syntax out of my brain. I can't recall any commands anymore 😂 | 13:33 |
| TheJulia | JayF: syntax purged and replaced via single-process thread | 13:33 |
| iurygregory | good morning ironic | 13:48 |
| dking | Any BMO folks around today? | 14:22 |
| opendevreview | Takashi Kajinami proposed openstack/ironic-tempest-plugin master: Add stable/2026.1 job https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/982243 | 14:33 |
| JayF | anandkaranubc: make sure you review stephen fin's answers starting at about 5:05a PST today ^^^ | 15:00 |
| opendevreview | Merged openstack/sushy master: Prevent double-wrapping timeout tuple in Connector _op https://review.opendev.org/c/openstack/sushy/+/982219 | 15:17 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Remove the SNMP hardware type and power interface https://review.opendev.org/c/openstack/ironic/+/982146 | 15:17 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Remove pysnmp and related dependencies https://review.opendev.org/c/openstack/ironic/+/982147 | 15:17 |
| JayF | iRMC change (not re-pushed here) is marked prio and ready to merge/review | 15:17 |
| JayF | this should be fixed but we'll see how devstack feels about that :D | 15:17 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: make target_datetime optional for set_bmc_clock step https://review.opendev.org/c/openstack/ironic/+/982261 | 15:21 |
| JayF | over -15k lines removal in those deprecations | 15:22 |
| * JayF pushing for 0 net LOC over OpenStack career :D | 15:22 | |
| JayF | Also; I'll note: used download-logs.sh + claude-code to troubleshoot SNMP CI. It's so ridiculously good at finding the issue. I had to keep it from writing a bad, naive fix, but I'm happy to do that since I didn't have to sort through those logs by hand | 15:23 |
| TheJulia | So, question, were we going to rip out ilo this cycle? | 16:19 |
| TheJulia | I feel like I mentally lost track of that one | 16:20 |
| JayF | I'm not sure | 16:20 |
| JayF | I am OK to do it | 16:20 |
| opendevreview | Abhishek Bongale proposed openstack/ironic-tempest-plugin master: Add conditional graphical console validation https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/982077 | 16:42 |
| opendevreview | Abhishek Bongale proposed openstack/ironic master: Limit interfaces to fake-graphical for testing https://review.opendev.org/c/openstack/ironic/+/982270 | 16:42 |
| opendevreview | Abhishek Bongale proposed openstack/ironic master: Limit interfaces to fake-graphical for testing https://review.opendev.org/c/openstack/ironic/+/982270 | 16:50 |
| cardoe | Would https://review.opendev.org/c/openstack/ironic/+/982261 be okay to backport? | 17:08 |
| JayF | It's a hardware compatibility fix, so I would default to yes | 17:09 |
| opendevreview | cid proposed openstack/ironic-specs master: A specification for a node power lock feature https://review.opendev.org/c/openstack/ironic-specs/+/974894 | 18:01 |
| JayF | TheJulia: re: iLO: in 2024.1 we had a release note indicating "ilo hardware types may be deprecated in the future". In 2026.1 (backported to 2025.2), we added an upgrade check to warn of ilo driver deprecation. | 18:16 |
| JayF | TheJulia: so from my PoV, we have a warning of deprecation in the latest slurp, so I think we are justified in removing it this cycle --- 2025.2->2026.2 is 2 releases, spanning one slurp | 18:17 |
| JayF | I will email the list and ask for people to scream if they disagree | 18:17 |
| TheJulia | k, Thanks. I saw we didn't explictly change the supported flag in the driver, but yeah, we've been messaging it for a while | 19:01 |
| cardoe | Soooo runbooks | 20:03 |
| cardoe | Working with https://review.opendev.org/c/openstack/ironic/+/982261 | 20:03 |
| cardoe | https://www.irccloud.com/pastebin/oafqmO1X/ | 20:03 |
| cardoe | But to let my operators use it. I've gotta call the runbook CUSTOM_RESET_BMC_CLOCK for example and add the trait CUSTOM_RESET_BMC_CLOCK. | 20:04 |
| cardoe | Then with automated runbook cleaning does that mean it's gonna be run every single time? | 20:05 |
| cardoe | So stupid idea.... extend runbooks to have a traits column which is a set and for the migration we add the name as only element in the set. Then maybe add a column for automated_cleaning? Cause we've said before runbook based deployment could be a think too. | 20:09 |
| JayF | yes and very very oh god no that's not how it works | 20:09 |
| JayF | I think you don't grok the shape of automated cleaning via runbook | 20:09 |
| cardoe | I don't. | 20:09 |
| JayF | https://static.opendev.org/docs/ironic/2025.2/admin/cleaning.html#configuring-automated-cleaning-with-runbooks | 20:10 |
| JayF | https://static.opendev.org/docs/ironic/2025.2/admin/cleaning.html#cleaning-steps is where it all starts | 20:10 |
| cardoe | Part of me wants to run the "reset_idrac" step but its very unclear what that actually does to the iDRAC... like am I gonna lose IP access? | 20:13 |
| JayF | dtantsur: TheJulia ^^ this is why I didn't want that step merged | 20:14 |
| JayF | cardoe: do not run it if you do not know the implications | 20:14 |
| JayF | cardoe: it's a factory reset and is breaky in some situations/environments | 20:14 |
| cardoe | Well it'd be helpful if we said what that did. | 20:14 |
| cardoe | There's also a "known_good_state" step which sounds warm and fuzzy but I checked and it internally calls the scary "reset_idrac" | 20:14 |
| cardoe | So that feels like a big foot gun just sitting there. | 20:15 |
| JayF | that's why I was against merging it initially, but RH had customers with specific use cases | 20:15 |
| cardoe | I mean it's fine to be in there but put some comments in the docs as to what its gonna do. And if its gonna lock me out put a warning. | 20:16 |
| cardoe | We've got a warning on the BIOS settings reset but not a warning on that. | 20:16 |
| JayF | we should perhaps add one then? | 20:16 |
| JayF | I don't go looking at the menu of our steps that often; especially out of band steps | 20:16 |
| cardoe | Yep. And I'm gonna ask to backport that. | 20:16 |
| * TheJulia reads and keeps reading becasue the discussion is going at full speed | 20:17 | |
| JayF | TheJulia: tl;dr cardoe went "what does the giant shiny red button do" w/r/t that reset_idrac step :D | 20:17 |
| * JayF dives over to stop him from mashing the button | 20:17 | |
| cardoe | I've got a person that's adding hardware in and coming to me every time she cannot do what she needs to do with Ironic. She doesn't have "admin" role or "system scope" btw. | 20:18 |
| TheJulia | Yeah, I don't remember what it does at the "RESET IT" mode, the softer mode afaik is just a "hey, restart with last known good config and refresh thyself" | 20:18 |
| TheJulia | ilo has some scarier knobs. Wait until we have control over firmware loading/control knobs! | 20:19 |
| cardoe | Well "known_good_state" step just calls "clear_job_queue" and "reset_idrac". And the docs for "known_good_state" just says "reset iDRAC to known good state". | 20:19 |
| cardoe | While "reset_idrac" says "reset the iDRAC" | 20:19 |
| TheJulia | reset might just be "reboot" | 20:19 |
| TheJulia | because, I remember the idrac folks were having to call that a LOT with wsman | 20:20 |
| cardoe | We've got a vendor command to reboot the iDRAC | 20:20 |
| * TheJulia shrugs and goes "its an idrac" | 20:20 | |
| cardoe | But this could be a case where we have "update_firmware" in the management interface and "firmware_update" in the firmware interface. | 20:20 |
| TheJulia | ahh, different modes! | 20:21 |
| cardoe | Anyway I still think my idea of changing runbooks to have the traits as a set and not having the name be the match-y thing is good. | 20:21 |
| TheJulia | for automated cleaning, maybe? | 20:22 |
| JayF | cardoe: Do you think you can run "N" runbooks as part of automated cleaning? Because you can't, and I think I would be -1 to that as a feature as it eliminates the value of the declarative cleaning: simplicity/easy to understand | 20:22 |
| cardoe | No. | 20:22 |
| JayF | I am really, really confused as to what the headache is? | 20:22 |
| JayF | you just want runbooks to map to N traits instead of having to add a trait to the nodes? | 20:22 |
| cardoe | I don't wanna put 7000 traits on every node? | 20:22 |
| JayF | oh, ohhhhhhhhhh | 20:23 |
| JayF | name = trait | 20:23 |
| JayF | so N runbooks = N traits | 20:23 |
| JayF | no way around it | 20:23 |
| TheJulia | I'm super confused | 20:23 |
| cardoe | CUSTOM_RESET_JOBS, CUSTOM_RESET_BMC_CLOCK, CUSTOM_RESET_DELL_RAID, CUSTOM_DELL_VENDOR_ACTION_X | 20:23 |
| JayF | TheJulia: right now, runbook name == a trait name (e.g. CUSTOM_BLAH_BLAH) | 20:23 |
| TheJulia | Computer, activate the Emergency Software Engineer! | 20:23 |
| cardoe | I just wanna put CUSTOM_IDRAC on those boxes | 20:23 |
| JayF | so if you want multiple runbooks to be valid for a node, you have to add the names of those runbooks to node traits | 20:23 |
| cardoe | Yes. | 20:23 |
| JayF | which has an ACTUAL PERFORMANCE IMPACT in scheduling | 20:23 |
| JayF | f* | 20:23 |
| cardoe | Yes | 20:23 |
| TheJulia | "Please state your user story" | 20:24 |
| cardoe | ^^^^ | 20:24 |
| cardoe | performance! | 20:24 |
| JayF | "I don't want to transmit over 9000 traits everytime nova syncs up" | 20:24 |
| cardoe | I'm watching nova and placement lose their minds | 20:24 |
| JayF | cardoe: +1 to the idea generally, we'd need to spec it | 20:24 |
| cardoe | yes | 20:24 |
| JayF | cardoe: if you wanna add it to PTG, cid might have opinions too since he wrote it originally | 20:24 |
| TheJulia | ++ | 20:24 |
| JayF | just add a column: traits: list(traits), migration sets trait[0]==name, future only enforces that it has a properly formatted trait, and match on that, not name | 20:25 |
| opendevreview | Julia Kreger proposed openstack/sushy master: WIP: Add TLS configuration support to sushy https://review.opendev.org/c/openstack/sushy/+/982309 | 20:31 |
| cardoe | Now make managing my certificate via ironic work nicely. | 20:32 |
| TheJulia | please elaborate on this user story request | 20:34 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Add global TLS configuration for API service https://review.opendev.org/c/openstack/ironic/+/982312 | 21:04 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Add TLS configuration for agent communications https://review.opendev.org/c/openstack/ironic/+/982313 | 21:04 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP: Add TLS configuration support to Redfish driver https://review.opendev.org/c/openstack/ironic/+/982314 | 21:04 |
| TheJulia | A lot of that is more than anything, discussion pointes to circle disucssions around | 21:10 |
| opendevreview | Harald Jensås proposed openstack/networking-generic-switch master: arista: Added ERROR_MSG_PATTERNS https://review.opendev.org/c/openstack/networking-generic-switch/+/982339 | 22:11 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Optimize L2VNI trunk manager with caching and DRY https://review.opendev.org/c/openstack/networking-baremetal/+/981650 | 22:33 |
| opendevreview | Harald Jensås proposed openstack/networking-baremetal master: Remove redundant IDL table existence checks https://review.opendev.org/c/openstack/networking-baremetal/+/982342 | 22:57 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!