| jrm | Hi. I'm doing some work the FreeBSD cloud-init-devel port/package. Is there a simple way to tell the cloud-init build tools to install the configuration files (cloud.cfg, 05_logging.cfg, and 99_freebsd.cfg) with .sample suffixes? I'll skip the details, but our package tools have some logic to 'do the right thing' with configuration files when they have a .sample suffix. | 00:00 |
|---|---|---|
| meena | jrm: that would be, if anywhere, under setup.py | 00:03 |
| meena | setuptools is just directly copying them. | 00:03 |
| Guest93 | meena: this is log from a "fresh boot" of a debian image: https://0bin.net/paste/oJ+LERXt#YH9NPj9ICeO-pyA6srBAiCvoEgRB/V69pRuIUWNVLEL | 00:04 |
| meena | so, unless setuptools has some hook, you'd have to patch setup.py to do the right thing… | 00:04 |
| jrm | meena: That seems to be the case for 05_logging.cfg, and 99_freebsd.cfg, but cloud.cfg is being generated directly to the staging dir. It's challenging with our USE_PYTHON=autoplist. | 00:05 |
| meena | Guest93: so, it's correctly identified, but can't find the datasource | 00:05 |
| Guest93 | Yea that's what I thought but I went through and updated the source and had it output the generated datasource; I copy/pasted it and cat'd it, it's definitely there, that's why I'm confused on if I'm missing something | 00:06 |
| Minghe | Hello, I've opened a new PR on the main to add mariner OS support on cloud-init. Can anyone review it and approve it pipeline testing since this is my first time contribute | 00:10 |
| Minghe | PR link: https://github.com/canonical/cloud-init/pull/1780 | 00:10 |
| -ubottu:#cloud-init- Pull 1780 in canonical/cloud-init "add mariner support" [Open] | 00:10 | |
| meena | Guest93: what does the user-data look like? | 00:10 |
| Guest93 | No userdata, only metadata currently | 00:11 |
| Guest93 | I think you're right about it not being able to find the data source. I've modified __init__.py and it only seems to find the DataSourceNone.py | 00:14 |
| Guest93 | But, the weird thing is that my data source is in the same dir | 00:15 |
| * meena hasn't written any data sources yet | 00:16 | |
| meena | Minghe: i would ping people who were active on reviewing other sources, or just anyone here with an @ hat. | 00:17 |
| * meena → bed. | 00:17 | |
| Minghe | got it | 00:48 |
| Minghe | falcojr would you able to review my PR? | 00:48 |
| === janitha76 is now known as janitha7 | ||
| falcojr | Minghe: yep! Will do. I saw some back and forth on that PR so I was waiting until the other comments were finished | 01:16 |
| Minghe | falcojr Thank you! | 01:43 |
| === gstoeh is now known as carael | ||
| carael | hello | 13:33 |
| meena | Hello carael ! | 13:35 |
| carael | how are you ? | 13:35 |
| * meena has added 28 lines of code, and now 28 tests are failing | 13:36 | |
| meena | perfect ratio, imo. | 13:36 |
| carael | is that the place to learn cloud init ? | 13:36 |
| meena | carael: it is. I'm a developer, of sorts, but I almost exclusively work on BSD support. | 13:39 |
| carael | great , nice to meet you | 13:40 |
| carael | where r u from then ? | 13:40 |
| meena | That is a very difficult question to answer. | 13:49 |
| meena | hrm, this is now failing tests in somewhat unexpected places, because initializing BSDNetworking calls subp(['ifconfig', '-a']) and lots of unrelated things are checking for one subp call exactly, and if they're getting an additional one, they don't like that | 15:21 |
| meena | so I'm gonna do some…thing… | 15:21 |
| meena | wow. down to 1 failure. | 15:32 |
| meena | someone needs to explain to me the purpose is of try_set_link_up() | 15:36 |
| meena | First off: I'm pretty, and sure, BSD can do that… but also: What's the point? | 15:37 |
| meena | blackboxsw, falcojr — i think https://github.com/canonical/cloud-init/pull/1779 is pretty solid now to get some eyes from y'all. A quality-check, before I proceed, and maybe some directions what to tackle next would be greatly appreciated. I should do some of my own: make sure this thing actually does what it claims | 15:50 |
| -ubottu:#cloud-init- Pull 1779 in canonical/cloud-init "Net: add BSD ifconfig(8) parser" [Open] | 15:50 | |
| falcojr | I'll take a look today | 15:51 |
| meena | but I have a feeling, if I try this, I'll get a lot less stray failed reads on /sys | 15:52 |
| meena | hrm, i wonder what good my caching is, if I don't store an Ifconfig() object… | 15:54 |
| meena | funny FreeBSD's cloud-init *could* run on more platforms, thanks to not using dmidecode… | 16:04 |
| * meena should go pick up her child… oops | 16:11 | |
| Guest93 | I'm testing a custom datasource and I noticed that the NIC name is different; installer has it default to ens3, but the NIC on the system is ens4. So, after cloud-init runs it doesnt bring the interface online. This normal type of behavior? | 18:09 |
| Guest93 | I know it's related to cloud-init because once I installed cloud-init the issue started to crop up, previously (no cloud-init) there was no issue. | 18:10 |
| falcojr | Guest93: it depends on what you mean by custom datasource. It's the datasource's responsibility to feed the networking information to cloud-init. | 18:15 |
| Guest93 | Ah, yea currently all it does is EphemeralDHCP to get access to the datasource | 18:16 |
| johnjaye | does cloud-init disable the grub selection menu or is that another app that does it? | 18:19 |
| falcojr | johnjaye: cloud-init doesn't do that | 18:21 |
| johnjaye | ok. i've been tinkering with debian cloud images trying to understand how they work. i noticed the nocloud image is the only one that allows editing of the grub menu | 18:22 |
| meena | johnjaye: chances are that the grub timeout has been set to 0 | 18:38 |
| falcojr | meena: Can you add your ifconfig output to the PR? I'm having some issues with the parsing code, but linux may output things differently | 19:13 |
| johnjaye | ah ok | 19:50 |
| meena | falcojr: i just used some of the existing ones | 19:52 |
| meena | falcojr: but i can give you my machines' output as examples | 19:52 |
| falcojr | yep, that's what I mean | 19:53 |
| meena | falcojr: https://gist.github.com/ea7ccf925a5a77cd834756867a812bab my local VM. a server running my websites: https://gist.github.com/5ee247db774f2b7d4250a64173535e98 — a local openbsd vm: https://gist.github.com/ae7e543260a77a691e6865358d2788d3 | 19:56 |
| meena | i just found something I won't be able to parse correctly! | 19:58 |
| meena | a tun device. | 19:58 |
| meena | https://gist.github.com/fef17df31318603b937b27d0dce1b01f | 19:59 |
| meena | falcojr: if my headache passes, and I manage to replace IPv6 and IPv4 addresses with RFC example addresses in the bunch of pastes i got from people, I can add those to our assets for broader testing | 20:05 |
| meena | right, let's build a package and see if cloud-init can get this machine configured! | 20:07 |
| meena | I mean, it's DHCP, i really hope it can lol | 20:07 |
| meena | (how do i pass some *other* network config via NoCloud?) | 20:08 |
| blackboxsw | meena: you can provide /var/lib/cloud/seed/nocloud-net/network-config (version 1 or version 2) if your meta-data and user-data files live there too. Oherwise you can provide a file in /etc/cloud/cloud.cfg.d/99-mynetwork.cfg with network: directives . `network-config=<Base64 encoded YAML config string>` can be provided on the kernel cmdline too if you have access to inject that config | 20:16 |
| meena | blackboxsw: I haven't fixed up FreeBSD to read the kernel line yet, and also: I have no idea how to pass a kernel line thru libvirt/qemu to FreeBSD | 20:17 |
| meena | blackboxsw: how come it can't just be done via the same ISO that the user-data/meta-data is passed in by? | 20:18 |
| blackboxsw | meena: you can provide the network-config file in the same directory or ISO that you pass it correct. Sorry lost context on how you are presenting a NoCloud source to the instance | 20:19 |
| blackboxsw | NoCloud datasource will opt to use the network-config file if present ... NoCloud also sources a 'vendor-data' file if you wanted optional vendor-data that could be overriddedn by user-data if you had some constomizations you were testing | 20:20 |
| blackboxsw | the required vs optional files are here https://github.com/canonical/cloud-init/blob/main/cloudinit/sources/DataSourceNoCloud.py#L88-L91 | 20:21 |
| meena | blackboxsw: i'm not quite there yet;) | 20:21 |
| meena | still just baby steps | 20:21 |
| blackboxsw | yeah it's hard to boil an ocean :) | 20:21 |
| meena | But I'm quite impressed how well my code seems to integrate so far… | 20:22 |
| meena | I think it should be really easy to rip out half of cloudinit.net in the next few weeks, and stick it into networking, with very minimal overhead given what Ifconfig/Ifstate already do | 20:23 |
| blackboxsw | you just invoked mi-ni-mal. meena are you tracking of current progress toward your goals and initial timing estimates in a public forum/page/hackmd etc? | 20:26 |
| meena | blackboxsw: not public, but I'm behind | 20:26 |
| blackboxsw | no worries. I was just wondering and expect that you'll continue communicating when things need attention | 20:27 |
| blackboxsw | and thanks | 20:27 |
| meena | chrony should add timesyncd to this table https://chrony.tuxfamily.org/comparison.html | 21:16 |
| meena | do lots of cloud providers use the rename feature of network devices? | 22:19 |
| meena | it's… terribly inconvenient… | 22:19 |
| meena | i'm gonna be importing re into every file i touch, if no one stops me | 22:53 |
| meena | https://gist.github.com/63a2163a644910128ebf463807a9cd14 | 22:53 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!