| arne_wiebalck | Good morning, Ironic! | 07:37 |
|---|---|---|
| rpittau | good morning ironic! o/ | 08:17 |
| janders | hey arne_wiebalck rpittau and Ironic o/ | 09:37 |
| arozman | Hello Ironic! | 09:55 |
| kubajj | Good morning arne_wiebalck, rpittau, janders, arozman and everybody! | 09:56 |
| janders | hey kubajj o/ | 09:59 |
| arne_wiebalck | hey kubajj o/ | 10:10 |
| opendevreview | Jacob Anders proposed openstack/sushy stable/zed: Fix setting boot related attributes https://review.opendev.org/c/openstack/sushy/+/865388 | 10:37 |
| opendevreview | Jacob Anders proposed openstack/sushy stable/yoga: Fix setting boot related attributes https://review.opendev.org/c/openstack/sushy/+/865411 | 10:46 |
| arne_wiebalck | TheJulia: for node history, rather than only logging errors (as I think it is now), would it make sense to also log state transitions (or at least have this as an option)? When debugging nodes, we usually know they are in error, and it would be good to know what the node has seen before in terms of transitions. | 13:01 |
| dtantsur | does anyone remember our logic behind not adding a transition: clean failed -> (clean) -> cleaning? | 13:59 |
| dtantsur | makes life harder for us in metal3 | 14:00 |
| dtantsur | (to say nothing that it makes it much easier to skip cleaning) | 14:06 |
| kubajj | dtantsur: what kind of api calls do we need for the inventory? | 14:33 |
| * dtantsur brain powering on | 14:33 | |
| dtantsur | kubajj: you mean, which calls do we need to add to Ironic API? or a step before that? | 14:33 |
| kubajj | dtantsur: the last step of the three, so Ironic API. Or do you think that I need something more for the second one? | 14:34 |
| dtantsur | kubajj: okay, so unless anybody here have better ideas, we need GET /v1/nodes/<NODE ID>/inventory that just returns a JSON with both "inventory" and "plugin_data" (or how did we call it?) | 14:36 |
| dtantsur | kubajj: now, I've realized we skipped one thing that inspector has: the backend for introspection data is actually configurable: database or swift | 14:36 |
| dtantsur | https://opendev.org/openstack/ironic-inspector/src/branch/master/ironic_inspector/plugins/introspection_data.py | 14:37 |
| dtantsur | or even no storage (for people who don't care) | 14:37 |
| TheJulia | dtantsur: no, I say add it | 14:45 |
| TheJulia | arne_wiebalck: I don't see why not add state transitions | 14:45 |
| * TheJulia powers up the coffee maker | 14:45 | |
| JayF | If we change that now, what do we say to the years that other operators have wtfed while running manage/provide over and over again | 14:55 |
| JayF | 😅 | 14:55 |
| JayF | That's a joke, if it's not clear | 14:56 |
| dtantsur | :D | 14:59 |
| * dtantsur has discovered that baremetal-operator does 'clean failed' -> 'manageable' and leaves the node there | 15:00 | |
| kubajj | dtantsur: should we add that before implementing the API, or after that? | 15:09 |
| dtantsur | kubajj: I don't have a really strong opinion, so up to you. Doing it before the API seems somewhat more logical. | 15:10 |
| kubajj | dtantsur: so is it connected to the change that is active now? | 15:13 |
| kubajj | I see that all the options just have a get and save method | 15:13 |
| dtantsur | kubajj: it could be a part of the active change, can be separate | 15:13 |
| dtantsur | yeah, what inspector has now is pluggable storage backends | 15:14 |
| dtantsur | set it to "none", you get no data stored, set to "swift" - it goes to OpenStack Swift (object storage) | 15:14 |
| dtantsur | if you find it harder to wrap your head around this new requirement, feel free to finish whatever is already planned | 15:15 |
| jrosser | i'm having a bit of difficulty adding rootpwd to kernel_append_params - do i need some escaping? https://paste.opendev.org/show/bUeLt0SMo6Kbgezk3Uz4/ | 15:22 |
| kubajj | dtanstur: I can look into it. Btw, I updated the test in the active change as you suggested. | 15:23 |
| dtantsur | jrosser: yeah, some escaping of $ is needed, I don't remember how exactly | 15:44 |
| jrosser | dtantsur: https://opendev.org/openstack/oslo.config/commit/df5e5db191c13198a12b3eb24f0eda603e3324d8 | 16:15 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP: Get conductor metric data https://review.opendev.org/c/openstack/ironic/+/865447 | 16:25 |
| TheJulia | so ^^^ might not work, but in kind of toying with the idea, and the realization we really have no good way of doing timer based metrics, I'm wondering if we get a timer metric, if we should also "count" the occurrence of the metric, which would get us a ton of metrics "for free" | 16:27 |
| TheJulia | if anyone has thoughts on adding that to ironic-lib, lmk ^ | 16:27 |
| rpittau | good night o/ | 17:13 |
| opendevreview | Julia Kreger proposed openstack/ironic master: WIP: Get conductor metric data https://review.opendev.org/c/openstack/ironic/+/865447 | 21:14 |
| janders | good morning Ironic o/ | 21:14 |
| janders | I'm trying to backport https://review.opendev.org/c/openstack/sushy/+/856597 all the way to Wallaby (and obviously to all the releases in-between). Zed and Yoga went smooth. From xena "downwards" auto cherrypick isn't working - due to lack of https://opendev.org/openstack/sushy/commit/519e54b1e9ace93deabf7140f9dff868a8e41de0 in the codebase | 21:16 |
| janders | JayF I suspect this is because the commit quoted is a complex change and not a bugfix but an improvement - hence it deliberately hasn't been backported (and for good reason). But I wanted to confirm with other folks before making assumptions | 21:17 |
| janders | Also - what do I do in such situation (was lucky enough it hasn't happened to me yet) - do I essentially rewrite https://review.opendev.org/c/openstack/sushy/+/856597 enough to match the (different) Xena/Wallaby codebase as a part of merging the code after a failed auto-cherrypick? | 21:18 |
| janders | (my default expectation) | 21:19 |
| TheJulia | if memory serves, the enum changes were a major change, your going to have to rebase without it I think | 21:22 |
| TheJulia | so yeah, as you put it "essentially rewrite" but use the same change id if you can so it can be appropriately tracked | 21:23 |
| janders | thank you TheJulia | 21:23 |
| TheJulia | you can edit on the cherry-pick and work to reconcile from there | 21:25 |
| janders | yeah reading through the ENUM change that may actually be not too bad | 21:25 |
| TheJulia | iurygregory: hey would you have a chance to chat say on Monday regarding ironic-prometheus-exporter? | 21:39 |
| iurygregory | TheJulia, yes | 22:16 |
| janders | TheJulia as I am working on the rebase of the Lenovo-fixing patch I wonder - once I'm done with Xena, can I backport the Xena backport to Wallaby (instead of backporting the master change)? | 23:54 |
| janders | (I'd expect it will just auto-cherrypick correctly with no manual interaction) | 23:54 |
| janders | just not sure if this is "proper" from the process angle, only even backported commits from master to branches | 23:54 |
| janders | I suppose change ID may be the same for all so it may not even be visible, but asking just in case | 23:55 |
| TheJulia | You are always supposed to cherry pick branch to branch | 23:57 |
| TheJulia | Not master to branch | 23:57 |
| TheJulia | Beyond the first hop | 23:57 |
| TheJulia | Doing that, allows you to accumulate the rebases understanding as you go and lessens the single point in time cost to cherry pick to older branches | 23:58 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!