| NobodyCam | morning janders o/ | 00:10 |
|---|---|---|
| TheJulia | o/ janders | 00:32 |
| TheJulia | how goes it? | 00:32 |
| janders | hey NobodyCam TheJulia o/ | 00:38 |
| janders | bit hectic - 4.10 code freeze in 2 days, but we're managing! :) | 00:39 |
| opendevreview | Vanou Ishii proposed openstack/ironic master: Fix failure of create_vfat_image function https://review.opendev.org/c/openstack/ironic/+/826576 | 03:09 |
| opendevreview | Vanou Ishii proposed openstack/ironic master: Fix failure of create_vfat_image function https://review.opendev.org/c/openstack/ironic/+/826576 | 03:55 |
| arne_wiebalck | Good morning, Ironic! | 07:28 |
| arne_wiebalck | dtantsur: after some back and forth, https://review.opendev.org/c/openstack/ironic-python-agent/+/826026 is back in the (Zuul) game ... would be great if we could get this in, then do i) the refactor and ii) the backports (if we decide to do it) | 08:11 |
| rpittau | good morning ironic! o/ | 08:26 |
| rpittau | arne_wiebalck: I had a look and it looks ok, also as backport-candidate, and I 100% agree with dtantsur about the extensions.image import, need to revisit something there | 09:13 |
| arne_wiebalck | rpittau: thanks | 09:51 |
| arne_wiebalck | rpittau: re extensions.image, I was thinking you would notice this after you have done all the efi/raid refactoring | 09:52 |
| arne_wiebalck | rpittau: I propose to do this in a followup | 09:52 |
| rpittau | heh I have to get back to that, this week is a bit busy :/ | 10:14 |
| dtantsur | morning folks | 10:27 |
| dtantsur | arne_wiebalck: re https://storyboard.openstack.org/#!/story/2009805: batching is what many ironic consumers do, I'm aware of metal3 and tripleo. | 10:28 |
| dtantsur | so yeah, totally +1 on that | 10:28 |
| dtantsur | (oh, and ironic-inspector has batching too) | 10:29 |
| dtantsur | now, failing vs delaying is a good question, and may even be an option for operators | 10:29 |
| arne_wiebalck | dtantsur: thanks for checking | 10:29 |
| dtantsur | arne_wiebalck: the RFE as it is written now needs technical details or even a spec | 10:30 |
| arne_wiebalck | dtantsur: yes | 10:30 |
| arne_wiebalck | dtantsur: if someone has ideas on *how* to do it, I may be able to a way to have it done :) | 10:31 |
| arne_wiebalck | *to find a way | 10:31 |
| arne_wiebalck | dtantsur: unless we come up with a plan how to do it until then, we could also briefly discuss this at the ptg | 10:51 |
| dtantsur | arne_wiebalck: we could add a new DB API, something like, nodes_in_status_count(['cleaning', 'clean wait']) | 10:52 |
| arne_wiebalck | dtantsur: so, upon a new clean request, we would call this and then fail or stall (somehow) ? | 10:56 |
| iurygregory | good morning Ironic | 10:58 |
| arne_wiebalck | hey iurygregory o/ | 10:58 |
| iurygregory | arne_wiebalck, o/ | 10:58 |
| rpittau | hey iurygregory dtantsur :) | 10:59 |
| dtantsur | arne_wiebalck: yeah. how to stall is a good question. | 11:00 |
| dtantsur | arne_wiebalck: we *might* need new state machines states. like 'clean delayed' or something | 11:00 |
| arne_wiebalck | dtantsur: what do clients do when they get a 503 with a retry header? | 11:04 |
| * arne_wiebalck is trying to outsource the problem :) | 11:04 | |
| dtantsur | arne_wiebalck: depends on the client? I don't think it's a good fit here | 11:04 |
| arne_wiebalck | dtantsur: ok | 11:04 |
| dtantsur | cleaning takes from minutes to hours, no client will retry that long | 11:04 |
| dtantsur | we'll also need to solve the issue with automated cleaning | 11:05 |
| dtantsur | we cannot really refuse deprovisioning | 11:05 |
| arne_wiebalck | right, we need to distinguish these | 11:05 |
| dtantsur | this is why I'm thinking of a new state | 11:06 |
| dtantsur | a node can be "parked" in "clean delayed"/"inspect delayed" until it has a slot | 11:06 |
| arne_wiebalck | yes | 11:07 |
| dtantsur | which can be checked in a periodic task on conductor | 11:07 |
| dtantsur | it will also be easy for an operator to understand why their nodes are not booting into the ramdisk | 11:07 |
| arne_wiebalck | a state would also be sth which an operator would understand | 11:07 |
| arne_wiebalck | :-D | 11:07 |
| arne_wiebalck | I will add our thoughts to the story | 11:08 |
| dtantsur | arne_wiebalck: another question: do we want to treat all booting steps (cleaning, deploy, inspection) together or separately? | 11:09 |
| arne_wiebalck | dtantsur: the initial motivation was to prevent irreversible damage | 11:10 |
| arne_wiebalck | dtantsur: therefore cleaning came to mind first, deploying is equally "dangerous", inspection should be less damaging I would think | 11:11 |
| arne_wiebalck | dtantsur: along those lines, what about deleting nodes from Ironic? | 11:12 |
| dtantsur | please elaborate, wdym by damage? | 11:12 |
| arne_wiebalck | dtantsur: remove instances, erase disks, ... | 11:12 |
| * arne_wiebalck goes for a quick lunch | 11:12 | |
| dtantsur | arne_wiebalck: I'm not sure I'm following. even with batching, the damage will still be done. | 11:12 |
| dtantsur | I was under assumptions you're trying to prevent a lot of PXE bootings from saturating your network | 11:13 |
| arne_wiebalck | dtantsur: right, there will still be damage, but it will be limited or can be stopped; if Ironic accepts 10k clean reqs and boots 10k nodes into cleaning, all instances could be gone in minutes; if you batch this into 100 max at any moment, you will have a chance to step in and stop it | 11:32 |
| dtantsur | hmm, interesting | 11:32 |
| arne_wiebalck | dtantsur: not saturating or limiting the network is an additional selling point I have not thought about :) | 11:34 |
| *** sshnaidm|afk is now known as sshnaidm | 11:35 | |
| opendevreview | Ameya Raut proposed openstack/ironic-tempest-plugin master: WIP : Add iDRAC management cleaning steps tests https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/826646 | 11:52 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Enable epel repository only when needed https://review.opendev.org/c/openstack/bifrost/+/826442 | 11:58 |
| opendevreview | Riccardo Pittau proposed openstack/bifrost master: Enable epel repository only when needed https://review.opendev.org/c/openstack/bifrost/+/826442 | 13:40 |
| dtantsur | FYI https://review.opendev.org/c/openstack/glance/+/826684 | 14:28 |
| TheJulia | ++ | 14:30 |
| opendevreview | Aija Jauntēva proposed openstack/ironic master: Fix DevStack plugin ipxe-snponly-x86_64.efi name https://review.opendev.org/c/openstack/ironic/+/826694 | 15:02 |
| opendevreview | Ameya Raut proposed openstack/ironic-tempest-plugin master: Add iDRAC management cleaning steps tests https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/826646 | 15:03 |
| arne_wiebalck | dtantsur: if you could give https://review.opendev.org/c/openstack/ironic-python-agent/+/826026 another thumps-up, I would start looking into backports (and refactor), so that we can finally close that chapter :) | 15:30 |
| dtantsur | looking | 15:32 |
| dtantsur | yep, +A | 15:33 |
| arne_wiebalck | thanks! | 15:34 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: [WIP] Explicit parameter to distinguish partition/whole-disk images https://review.opendev.org/c/openstack/ironic/+/825305 | 15:37 |
| TheJulia | so random thought, w/r/t power arch. I wonder if it might be time to deprecate PReP partition support | 16:21 |
| dtantsur | TheJulia: is it not needed any more? | 16:31 |
| TheJulia | I don't think so, I think the latest generation no longer supports it | 16:31 |
| TheJulia | I *think* | 16:31 |
| dtantsur | would be good to get a confirmation | 16:32 |
| TheJulia | yeah | 16:32 |
| TheJulia | oh where oh where is that email address | 16:33 |
| dtantsur | I definitely don't mind removing so much code that none of us can even test | 16:36 |
| rpittau | good night! o/ | 17:06 |
| TheJulia | dtantsur, I've copied you on an email to mike who authored it originally | 17:07 |
| dtantsur | yep, thanks! | 17:09 |
| dtantsur | see you tomorrow folks o/ | 17:09 |
| opendevreview | Merged openstack/ironic-python-agent master: SoftwareRAID: Use efibootmgr (and drop grub2-install) https://review.opendev.org/c/openstack/ironic-python-agent/+/826026 | 17:13 |
| *** pmannidi is now known as pmannidi|Gone | 17:29 | |
| arne_wiebalck | \o/ (this should hopefully mark the end of grub2-install for s/w RAID and makes it use the same paths as any other UEFI deployment) | 17:33 |
| TheJulia | the world will be a better place | 17:35 |
| arne_wiebalck | :-D | 17:37 |
| iurygregory | arne_wiebalck, \o/ | 17:48 |
| arne_wiebalck | iurygregory: :) | 17:48 |
| iurygregory | the wolrd will be a better place and we are one step closer to take over the galaxy :D | 17:48 |
| arne_wiebalck | bye everyone, see you tomorrow o/ | 17:49 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: WIP: An idea for rbac positive/negative testing https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/819165 | 17:51 |
| iurygregory | bye arne_wiebalck o/ | 17:51 |
| TheJulia | stevebaker[m]: hjensas: w/r/t that bz yesterday, it magically worked this morning... | 18:12 |
| *** sshnaidm is now known as sshnaidm|afk | 18:13 | |
| hjensas | TheJulia: that's good news for sanity. | 18:23 |
| TheJulia | hjensas: yeah | 18:24 |
| TheJulia | unfortunately it sunk two days of my brain thinking I was absolutely starting to loose it | 18:25 |
| opendevreview | Merged openstack/ironic master: Fix DevStack plugin ipxe-snponly-x86_64.efi name https://review.opendev.org/c/openstack/ironic/+/826694 | 18:26 |
| hjensas | TheJulia: feel free to redirect the rear stuff to me in case they raise something. I know you have a lot on your plate. | 18:31 |
| TheJulia | hjensas: thanks, I looked at it mainly because I remembered talking to them when they developed the feature originally | 18:33 |
| TheJulia | and kind of kept going "what?!?" | 18:33 |
| TheJulia | stevebaker[m]: I'm starting to think ddt.... | 19:04 |
| TheJulia | stevebaker[m]: which would allow us to fork lift a lot out of ironic today, into ironic-tempest-plugin and remodel it I guess | 19:20 |
| TheJulia | the question ebcomes how to model "this should work, but not that and that, but then this other case should also work" | 19:20 |
| stevebaker[m] | Yeah ddt is great if you know up front that it can implement your most complex test | 19:33 |
| TheJulia | the whole unknown is how to represent that matrix I guess | 19:41 |
| NobodyCam | Good AfterNoon Ironic'ers | 20:28 |
| NobodyCam | TheJulia: I wanted to check with you on https://review.opendev.org/c/openstack/nova/+/813897 ... looks to be DOA? | 20:29 |
| iurygregory | good afternoon NobodyCam o/ | 20:31 |
| NobodyCam | hey hey iurygregory :) happy almost Friday | 20:32 |
| iurygregory | happy almost Friday \o/ | 20:32 |
| NobodyCam | heheheh | 20:32 |
| NobodyCam | glad to back? saw some great photos on "the" twitter | 20:33 |
| iurygregory | the PTO was quite good :D | 20:39 |
| iurygregory | energy restored =) | 20:39 |
| NobodyCam | \o/ | 20:41 |
| TheJulia | NobodyCam: ENOTIME | 21:12 |
| NobodyCam | TheJulia: +++ completely understand, wanted to check | 21:13 |
| NobodyCam | Hope things are going well today! | 21:13 |
| TheJulia | lots of long discussions | 21:13 |
| NobodyCam | :) | 21:13 |
| TheJulia | NobodyCam: if you feel that would solve your issue, I would highly recommend commenting and the version you would love to see a fix in | 21:59 |
| NobodyCam | ++ | 22:00 |
| TheJulia | Granted, it may be an EM version, often these are the sort of bugs we just go ahead and backport through em versions to older versions we're supporting for long term support | 22:00 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!