mkrai_ | Hi cloud-init team. I am trying to deploy a node when an infiniband card with OpenStack. cloud-init fails complaining that the MAC address of the infiniband card is not know. OpenStack provides fake 6bytes MAC address in format mac[36:-14] + mac[51:]. | 08:02 |
---|---|---|
mkrai_ | Can someone please help me? | 08:02 |
meena | mkrai_: hello good morning | 10:12 |
mkrai_ | meena, Hi, good morning | 10:13 |
meena | n.b.: most cloud-init committers live somewhere across the north american continent. The help that i can provide is probably not going to be of the same quality. | 10:13 |
meena | mkrai_: can you show us the exact error message? | 10:14 |
mkrai_ | It didn't log any error, but I debugged it and found the error https://github.com/canonical/cloud-init/blob/master/cloudinit/sources/helpers/openstack.py#L677 | 10:16 |
mkrai_ | meena, ^ | 10:19 |
meena | mkrai_: that's very recent code, i *think* which version are you running? | 10:20 |
meena | nope, never mind, confusing it with… something else. | 10:20 |
mkrai_ | meena, same error but I am running 18.5 | 10:22 |
meena | oh, it actually is from there: https://github.com/canonical/cloud-init/pull/122 | 10:23 |
meena | this is where the discussion happens: https://github.com/canonical/cloud-init/pull/122#pullrequestreview-339327984 | 10:23 |
meena | mkrai_: do you, per chance, work with the author of that patch? | 10:24 |
meena | maybe you two should talk :P | 10:25 |
mkrai_ | meena, I am trying to reach him ;) | 10:26 |
meena | from their av at Github it would be "her" | 10:26 |
meena | anyway, it seems madhuri-rai07 hit a similar problem, and decided to deal with it outside of cloud-init. The question whether the problem is OpenStack or Cloud-Init is hard to say without access to such (expensive?) hardware… | 10:29 |
mkrai_ | meena, i am madhuri-rai07 ;) sorry i misunderstood | 12:03 |
meena | mkrai_: duuh, sorry 😅 | 12:44 |
meena | mkrai_: so, am i understanding you correctly, you've narrowed down the issue to: infiniband or, infiniband thru OpenStack? | 13:05 |
meena | or is the issue _also_ in Cloud-Init? | 13:06 |
Odd_Bloke | akik: Are you still looking for help with your default user question? | 15:03 |
akik | Odd_Bloke: yes, please | 15:06 |
akik | i had user section in my cloud-config but i removed it, and then i got that default centos account | 15:06 |
Odd_Bloke | akik: So I believe it is `users: - default` that is creating the default user (because generally people _do_ want a user created, even if they haven't explicitly specified it). Are you looking to create _no_ users, or to create only the users you specify? | 15:08 |
akik | Odd_Bloke: i'd like cloud-init not to create users | 15:09 |
Odd_Bloke | akik: If you're passing user-data, then I believe `users: []` would do the trick. | 15:10 |
akik | Odd_Bloke: oh right, thanks | 15:11 |
Odd_Bloke | If you want to do it via /etc/cloud, then I think `users: []` in a /etc/cloud/cloud.cfg.d/ snippet would do it too (though I haven't tested that). | 15:12 |
akik | Odd_Bloke: what would i remove from /etc/cloud/cloud.cfg to do the same? | 15:13 |
akik | i see the users: there | 15:13 |
akik | oh ok cloud.cfg.d might be better | 15:13 |
Odd_Bloke | akik: I think removing the users stanza would do it, but I'd recommend using a snippet so you aren't manually editing a config file that a package installed. | 15:13 |
Odd_Bloke | (Because on Debian/Ubuntu at least, if you change that file then you'll be prompted to resolve conflicts any time that config file changes on upgrade. The snippet doesn't have that issue.) | 15:14 |
akik | yes i'm using user-data via cloudsigma's api | 15:18 |
Odd_Bloke | OK, cool, then that's your best option. :) | 15:33 |
Odd_Bloke | rharper: https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1860046 is the RPi bug we were waiting for, FYI. | 17:25 |
ubot5 | Ubuntu bug 1860046 in cloud-init (Ubuntu) "Race to mount seed device" [Undecided,New] | 17:25 |
meena | Odd_Bloke: hello, welcome back, have you considered refactoring cloudinit/net today? | 17:56 |
Odd_Bloke | meena: I haven't, and as I'm (a) catching up from Friday, and (b) all the US folks are on vacation, I probably won't today I'm afraid. | 19:39 |
meena | Odd_Bloke: no stress… just (daily) nagging. | 19:46 |
Odd_Bloke | :) | 19:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!