| TheJulia | stevebaker[m]: so maybe members view only by default and lessees no access, and admins write?! | 00:05 |
|---|---|---|
| stevebaker[m] | TheJulia: that sounds reasonable to start | 00:30 |
| JayF | +1 | 01:11 |
| opendevreview | Doug Goldstein proposed openstack/ironic master: fix glance metadata layout https://review.opendev.org/c/openstack/ironic/+/942496 | 02:46 |
| rpittau | good morning ironic! happy friday! o/ | 07:50 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: Use bigger partition as work dir for metal3 job https://review.opendev.org/c/openstack/ironic/+/943374 | 08:31 |
| opendevreview | Riccardo Pittau proposed openstack/ironic master: [WIP] Use bigger partition as work dir for metal3 job https://review.opendev.org/c/openstack/ironic/+/943374 | 09:37 |
| adam-metal3 | rpittau, sorry to disturb you but I was wondering how necessary of cacheing the boot and vendor info within the pwoer_sync process of Ironic here : https://opendev.org/openstack/ironic/src/branch/master/ironic/conductor/manager.py#L4092 ? With deployments that have nodes north of 100 this will result sending 300 requests within 1 minute just for power syncing, is it really necessary to do the cacheing here?, so would anything break if I would | 09:43 |
| adam-metal3 | just remove this to chacing calls from the powr cycling ? | 09:43 |
| adam-metal3 | two* caching calls | 09:43 |
| adam-metal3 | I will do test to see if regular standalone ironic use-cases I have would break if I just remove these 2 cahing actions or not | 09:44 |
| rpittau | adam-metal3: IIRC that has been tested with a large deployment and it didn't give any issue | 09:45 |
| adam-metal3 | rpittau: It does not break, but I have received complaints from my stakeholders that reaching out to the same endpoint 3times within 1 minute x 100+ is seen as an unreasonable network trafic generation, I understand that by itself it is not a "good enough" upstream justification but I was wondering how ncessary it is to do these 2 cahce updating operations together with power syncing | 09:48 |
| adam-metal3 | I mean within the power status syncing itself | 09:48 |
| adam-metal3 | because AFAIK this vendor and boot info is cached during other operations also so here ti looks like they are called for the reason of "just to make sure it is up to date" | 09:50 |
| rpittau | adam-metal3: I don't remember the exact story behind it, but looking at it yes, it's done to be sure it's up to date in case operations are needed for specific vendors/drivers | 09:52 |
| rpittau | and well boot_mode really need to be correct in case of power status change | 09:53 |
| adam-metal3 | rpittau: thanks for confirming, I think I have an idea to modify this to improve this process, will let the community know if I have smething concrete | 09:58 |
| rpittau | np :) | 10:01 |
| vsaienko | please review https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/943529 marks BaremetalDriverAnsibleWholedisk as unstable | 10:26 |
| *** iurygregory__ is now known as iurygregory | 11:50 | |
| iurygregory | happy friday ironic | 11:50 |
| iurygregory | seems like iDRAC 10 changed a few things on the OemManager.ImportSystemConfiguration D: | 11:53 |
| opendevreview | Merged openstack/sushy-tools master: Redefine libvirt domain on Restart https://review.opendev.org/c/openstack/sushy-tools/+/943239 | 12:02 |
| opendevreview | Takashi Kajinami proposed openstack/ironic master: vncproxy: Use dedicated options for SSL https://review.opendev.org/c/openstack/ironic/+/943678 | 12:14 |
| opendevreview | Takashi Kajinami proposed openstack/ironic master: vncproxy: Use dedicated options for SSL https://review.opendev.org/c/openstack/ironic/+/943678 | 12:15 |
| tkajinam | ^^^ this probably needs some attention before rc1 | 12:15 |
| tkajinam | (this also made me aware the existing consoleproxy feature in nova lacks cafile option which I'll fix ... | 12:16 |
| iurygregory | tks tkajinam! | 12:17 |
| tkajinam | :-) | 12:18 |
| rpittau | an y available core for a quick approval to unblock CI? :) | 13:29 |
| rpittau | please check https://review.opendev.org/c/openstack/python-ironicclient/+/943584 and https://review.opendev.org/c/openstack/python-ironic-inspector-client/+/943588 | 13:29 |
| dtantsur | done | 13:34 |
| dtantsur | Should I be concerned that we've changed something that may break people? | 13:34 |
| dtantsur | tkajinam: oh, removal of sslutils can be a pretty big disaster | 13:35 |
| rpittau | dtantsur: I'm still trying to understand what happened there, we didn't change anything on the clients side that can justify that | 13:38 |
| dtantsur | A new version of osc-lib? | 13:38 |
| rpittau | mmmm maybe | 13:38 |
| rpittau | I will have to backport the fix to the new stable branches now.... :/ | 13:39 |
| dtantsur | le sigh | 13:40 |
| opendevreview | Merged openstack/python-ironicclient master: Fix bare metal info order in unit tests https://review.opendev.org/c/openstack/python-ironicclient/+/943584 | 13:45 |
| opendevreview | Merged openstack/python-ironic-inspector-client master: Fix bare metal info order in unit tests https://review.opendev.org/c/openstack/python-ironic-inspector-client/+/943588 | 13:46 |
| opendevreview | Riccardo Pittau proposed openstack/python-ironicclient stable/2025.1: Fix bare metal info order in unit tests https://review.opendev.org/c/openstack/python-ironicclient/+/943691 | 13:47 |
| opendevreview | Riccardo Pittau proposed openstack/python-ironic-inspector-client stable/2025.1: Fix bare metal info order in unit tests https://review.opendev.org/c/openstack/python-ironic-inspector-client/+/943692 | 13:47 |
| opendevreview | cid proposed openstack/ironic-tempest-plugin master: Test double encoding of error message https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/935740 | 14:43 |
| opendevreview | cid proposed openstack/ironic-python-agent master: Doc: Fix incorrect function in example code https://review.opendev.org/c/openstack/ironic-python-agent/+/943771 | 14:44 |
| opendevreview | Dr. Jens Harbott proposed openstack/ironic master: Revert "Add systemd provider for console containers" https://review.opendev.org/c/openstack/ironic/+/943798 | 15:09 |
| keekz | do nova aggregates work with ironic? i'm looking at https://docs.openstack.org/nova/latest/admin/aggregates.html .. i'd like to be able to put all the hosts in a specific rack in an aggregate or something similar, so that i can issue a build with that rack trait | 15:16 |
| cardoe | frickler: I just wanna fix glance usage in a backportable way. My follow on (which I had wanted to land in 2025.1 but now looks less likely) refactors Ironic to use the real object type throughout instead of thunking to a dict and passing a dict all around. But it's a bigger change and changes the tests a lot more. It fixes another half dozen bugs where we do our_dict.get('some_typo', 'default_value') which results in us | 15:22 |
| cardoe | always using the default. | 15:22 |
| cardoe | I'm also unsure of how those fixes will affect user usage because some of those typos have been around for 3+ releases. | 15:23 |
| frickler | cardoe: backporting is a good argument, ack | 15:48 |
| JayF | keekz: a lot of active ironic irc'ers are here at scale/openinfra days. Might be a good question for the mailing list in general though | 17:00 |
| keekz | ok, thanks for the heads up | 17:14 |
| opendevreview | Pierre Riteau proposed openstack/metalsmith master: Use releases.openstack.org instead of opendev.org https://review.opendev.org/c/openstack/metalsmith/+/943810 | 17:57 |
| cardoe | just a heads up, I'm on PTO next week. | 18:45 |
| TheJulia | iurygregory_: so I can remove the bulk config option the. ;) | 19:19 |
| vsaienko | TheJulia, cardoe: please review https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/943529 marks BaremetalDriverAnsibleWholedisk as unstable, with this patch chain with 3 patches at least was able to pass CI (not with first attempt, due to not related timeouts). | 19:21 |
| TheJulia | vsaienko: unstable on top of dnsmasq change? | 19:23 |
| vsaienko | what dbsmasq change, I'm sorry I'm not aware about that change | 19:25 |
| vsaienko | redfish fas failing with: Stderr: 'env: ‘ansible-playbook’: No such file or directory\n'. | 19:26 |
| vsaienko | not sure its related to dnsmasq | 19:26 |
| vsaienko | please check this bug https://bugs.launchpad.net/ironic/+bug/2101021 | 19:26 |
| TheJulia | Ack, thanks. I’m at lunch at SCaLE | 19:36 |
| TheJulia | The dnsmasq issues were it crashing :( | 19:39 |
| vsaienko | https://bugs.launchpad.net/ironic/+bug/2026757 this one? | 19:40 |
| vsaienko | I though that on CI dnsmasq is pinned to some earlier version https://github.com/openstack/ironic/blob/25a3dd076a0a8d3f4bbb5886252f6d08d78e33f9/devstack/lib/ironic#L3698 | 19:43 |
| opendevreview | Takashi Kajinami proposed openstack/ironic master: vncproxy: Use dedicated options for SSL https://review.opendev.org/c/openstack/ironic/+/943678 | 20:21 |
| opendevreview | Satoshi Shirosaka proposed openstack/ironic-python-agent-builder master: WIP Add Podman DIB for container-based cleaning https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/943348 | 20:24 |
| iurygregory_ | TheJulia, yeah I think we don't need to send https://opendev.org/openstack/sushy-oem-idrac/src/branch/master/sushy_oem_idrac/resources/manager/manager.py#L86 anymore (at least for iDRAC 10) | 20:26 |
| iurygregory_ | going to do some testing next week with the two machines I have access to figure out how to solve it | 20:27 |
| JayF | Just a thought of a use case we might want to consider; think about this: I have an entire datacenter of servers, thousands of servers, managed by ironic. The BMC network goes kaput, and the power status loop marks all of them maintenance. That's thousands of API calls to unset maintenance. | 20:32 |
| JayF | I don't really like the idea of a shape of a bulk-set/unset endpoint, but it might be worth thinking about how to auto-recover this, or allow an operator to rapidly fix somehow(?) | 20:33 |
| opendevreview | Satoshi Shirosaka proposed openstack/ironic-python-agent-builder master: WIP Add Podman DIB for container-based cleaning https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/943348 | 21:06 |
| opendevreview | Satoshi Shirosaka proposed openstack/ironic-python-agent-builder master: WIP Add Podman DIB for container-based cleaning https://review.opendev.org/c/openstack/ironic-python-agent-builder/+/943348 | 21:26 |
| opendevreview | Dr. Jens Harbott proposed openstack/ironic master: DNM: Revert some changes for testing https://review.opendev.org/c/openstack/ironic/+/943798 | 21:33 |
| *** winicius is now known as wncslln | 21:39 | |
| JayF | vsaienko: that pin broke for a couple weeks, which caused the real bad instability. | 22:11 |
| JayF | I'm on conference wifi dropping about 50% of my packets so I may not be able to help; but looking at 943798 what broke? Is there a bug somewhere? | 22:11 |
| frickler | JayF: that's my current guess for https://bugs.launchpad.net/ironic/+bug/2101021 which vsaienko mentioned. just going by correlation for now, no idea about cause except if it is only the bindep change, will test that next | 22:15 |
| JayF | will look if it decides to load :D | 22:15 |
| JayF | but if you're concerned about those, make sure to loop in stevebaker[m] -- he's been deep in those console patches and it'd be a shame if we had to revert them | 22:15 |
| JayF | (I assume the ping on gerrit might get lost in a sea of notifications) | 22:16 |
| JayF | ah, ansible driver :) Yeah I'm not going to be a huge amount of help there, but that's absolutely a wweird thing to break | 22:16 |
| opendevreview | Dr. Jens Harbott proposed openstack/ironic master: DNM: testing ironic-standalone-redfish job https://review.opendev.org/c/openstack/ironic/+/943827 | 22:17 |
| frickler | seems the failure is not happening on 943798, now let's see what ^^ does | 22:18 |
| frickler | JayF: yeah, I have no intention of actually reverting this, just trying to isolate the issue | 22:19 |
| JayF | just punch the 'gerrit bisect' button /s :D | 22:19 |
| JayF | I figured, but he also might have ideas too. And honestly I am always understanding of a "fix ci" revert when we can repush at fix | 22:19 |
| JayF | just ... perilously close to the release 😅 | 22:19 |
| stevebaker[m] | Devstack doesn't actually install those bindep.txt devstack packages so they are really only there for documentation. Those bindep changes can be backed out if necessary | 22:31 |
| TheJulia | iurygregory_: cool cool, I was more so thinking the feature in ironic | 23:39 |
| TheJulia | JayF: we've had auto-recover for that for... years and years. | 23:39 |
| TheJulia | (the thousands of nodes go into maintenance over power state check failing thing) | 23:40 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!