| [13:17] <minimal> with the recent rationalisation of module settings' names to use underscores rather than minuses should the names of modules in cloud.cfg.tmpl also standardise on using underscores? | |
| [13:18] <falcojr> minimal: yes, both are currently supported but we should really standardize on one | |
| [13:24] <minimal> perhaps I'll raise a PR to do this | |
| [13:26] <minimal> also I'm confused about gateway4 & gateway6 in network-config v2 - recently I noticed these were depricated so changed my YAML files to use "route" instead. However a recent PR appears to have removed the deprecated note from the docs (though I don't see which PR did it) | |
| [13:26] <minimal> so are gateway4/gateway6 deprecated or not for network-config v2? | |
| [13:28] <falcojr> we removed the deprecated note? It's still deprecated for v2. For v1, we changed the netplan implementation to use routes instead | |
| [13:29] <minimal> yupe, it was removed some time since Nov 21 (as I just did a "git pull" which last run then) | |
| [13:29] <minimal> doc/rtd/topics/network-config-format-v2.rst was updated to remove the deprecated note | |
| [13:30] <minimal> maybe because the previous note had a reference to "netplan#default-routes"? I'm using v2 with eni myself. | |
| [13:31] <falcojr> hmmm, I think that might've just been a mistake/oversight | |
| [13:50] <falcojr> I put up a PR to fix it. Thanks for letting us know! | |
| [14:04] <minimal> np, I have another related issue to investigate as I have found that using "to: default" does NOT work for network-config v2 "routes:" (at least for eni) | |
| [14:05] <minimal> the docs mention default but ~I had to use 0.0.0.0 to get it working | |
| [14:22] <Teridon> I'm having with cloud-init 22.4.2-0ubuntu0~20.04.1 . I defined 3 types of SSH host keys and certs in user-data. All 3 keys and certs ended up in /etc/ssh/, but in sshd_config, there is only one HostCertificate line for the RSA key. Did I do something wrong ? user-data excerpt : https://dpaste.org/EL03J cloud-init.log: https://dpaste.org/R1EhU | |
| [14:23] <Teridon> jeez. obviously. "I'm having *a problem* with ... " | |
| [14:28] <minimal> Teridon: I don't understand the problem, you specified 3 different certs in user_data and 3 HostCertificate lines were written | |
| [14:29] <falcojr> there's a single host certificate in /etc/ssh/sshd_config.d rather than all 3 | |
| [14:29] <minimal> ah, that it wrote 1 line, updating it 2 times | |
| [14:30] <falcojr> Teridon: Looks like we have a bug. Would you mind filing this at https://bugs.launchpad.net/cloud-init ? | |
| [14:31] <minimal> the sshd_config manpage entry for HostCertificate says "Specifies *a* file...", that's file singular | |
| [14:31] <Teridon> You can specify HostCertificate multiple times | |
| [14:31] <minimal> it's not clear if sshd_config can have multiple HostCertificate lines though | |
| [14:32] <minimal> ok, the manpage doesn't indicate that either way | |
| [14:32] <Teridon> yes it can. I do that normally on all my systems... and sshd presents the proper certificate depending on the key type you use to connect | |
| [14:33] <Teridon> I will file a bug, thx | |
| [14:35] <minimal> looks like the update_ssh_config_lines function only adds a line if it is not present and updates it otherwise, so doesn't cater for adding multiple such lines | |
| [14:35] <minimal> in cloudinit/ssh_util.py | |
| [15:27] <meena> https://github.com/canonical/cloud-init/compare/main...igalic:cloud-init:bsd/device_drivers?expand=1 i had to do this on my phone, because my laptop('s battery) died | |
| [15:28] <meena> minimal: netplan runs on alpine? | |
| [15:29] <meena> minimal: does netplan require dbus? | |
| [15:32] <meena> https://github.com/canonical/netplan/blob/main/meson.build#L28 can't tell if this is a joke | |
| [15:33] <falcojr> nah, I'm sure that find_program is searching the path...find is different | |
| [15:35] <meena> anyway, i can't tell from that meson file of systemd and dbus are hard dependencies | |
| [15:38] <meena> can't find netplan here, https://git.alpinelinux.org/aports/log/?qt=grep&q=netplan | |
| [15:43] <minimal> meena: no, I'm using network-config v2 with eni, not netplan | |
| [15:43] <minimal> what made you think I/Alpine was using netplan? | |
| [15:45] <meena> minimal: my misreading of what you said | |
| [15:46] <minimal> meena: there's no systemd on Alpine, someone tried to package it about 6 months ago and it didn't go down well... | |
| [15:47] <meena> yeah, just looking at netplan's code, looks like it's deeply interwoven with systemd | |
| [15:49] <minimal> meena: the issue with trying to get systemd on Alpine was two-fold, (1) the package submitter sort of indicated he was doing so to troll some Alpine people and (2) systemd people do not support musl and will accept any musl-related packages and so if an Alpine package was ever added it would have to carry around a large number of downstream patches forever | |
| [15:50] <minimal> s/will accept/will not accept/ | |
| [15:50] <minimal> s/packages/patches/ | |
| [15:50] <minimal> my typing is awful today lol | |
| [15:56] <meena> hands frozen? | |
| [15:57] * meena points at https://github.com/freebsd/freebsd-ports/tree/main/www/chromium/files | |
| [15:57] <minimal> nope, probably not enough/too much coffee ;-) | |
| [16:00] <meena> most of these patches are so trivial it's deeply frustrating | |
| [16:02] <meena> https://github.com/freebsd/freebsd-ports/blob/main/www/chromium/files/patch-chrome_browser_metrics_power_process__metrics__recorder__util.cc | |
| [16:03] <meena> AIX: chromium supported platform. BSD? nope | |
| [17:06] <Teridon> I tried to submit a bug at https://bugs.launchpad.net/cloud-init/+filebug but the site had some kind of error | |
| [17:07] <blackboxsw> checking Teridon an OOPS message ? | |
| [17:07] <Teridon> yes. Unfortunately I didn't get the code. I tried to go back to my bug form to resubmit but it was gone | |
| [17:07] <blackboxsw> that route looks like it works for me... checking to see if your bug got submitted | |
| [17:08] <blackboxsw> hrm latest bug I see is https://bugs.launchpad.net/cloud-init/+bug/1999042. so something didn't submit | |
| [17:08] -ubottu:#cloud-init- Launchpad bug 1999042 in cloud-init "In openEuler, distro.osfamily is incorrect" [Undecided, New] | |
| [17:09] <Teridon> let me try again I guess | |
| [17:14] <Teridon> Well I think I got the bug submitted (1999164) but I just failed to add an attachment (Error ID: OOPS-487354aacd8f1d4dbe05652042959d86) | |
| [17:42] <Teridon> got both my attachments done. I guess it was something with my proxy https://bugs.launchpad.net/cloud-init/+bug/1999164 | |
| [17:42] -ubottu:#cloud-init- Launchpad bug 1999164 in cloud-init "when multiple SSH host key certificates are defined, only one HostCertificate is referenced in sshd_config" [Undecided, New] | |
| [18:50] <minimal> falcojr: remember when I mentioned an issue a month ago where I was having problems on Alpine with the boot services and cloud-init.log all complained about distro being ubuntu and I eventually figured out it was a syntatically invalid /etc/cloud/cloud.cfg file? | |
| [18:52] <minimal> I had had another similar report on Alpine and it think it was introduced in cloud-init in a PR between 22.3 and 22.4 - the 2nd line of cloud.cfg.tmpl is a blank line and once the jinja templating occurs to create cloud.cfg that 2nd blank line becomes a blank 1st line, which yamllint says in a syntax error | |
| [18:52] <blackboxsw> do tell minimal: invalid YAML? | |
| [18:52] <blackboxsw> interesting. checking commit history there | |
| [18:54] <blackboxsw> wow so a blank line is a yaml error? | |
| [18:54] <blackboxsw> checking yamllint behavior | |
| [18:56] <blackboxsw> it's most likely this commit https://github.com/canonical/cloud-init/commit/cd2cca35a1 but hrm. | |
| [18:56] -ubottu:#cloud-init- Commit cd2cca3 in canonical/cloud-init "Create reference documentation for base config" | |
| [19:05] <minimal> blackboxsw: not simple a blank line, I believe the *1st* line of a valid YAML file cannot be blank | |
| [19:06] <minimal> yamllint says: | |
| [19:06] <minimal> 1:1 error too many blank lines (1 > 0) (empty-lines) | |
| [19:07] <blackboxsw> hrm.yet yaml.safe_load('\n\n1:2') seems to process fine :/ | |
| [19:07] <holmanb> looks like a configurable yamllint value | |
| [19:07] <holmanb> https://yamllint.readthedocs.io/en/stable/rules.html#module-yamllint.rules.empty_lines | |
| [19:09] <blackboxsw> I have no qualms with us dropping that leading new line. ... it's not needed for anything and as part of a jinja template it doesn't make sense to have the spacer anyway | |
| [19:09] <blackboxsw> as the file will get processed before rendering. | |
| [19:10] <minimal> yeah it's certainly technically not valid to have it | |
| [19:10] <blackboxsw> minimal: is there something else that is malformed in the jinja rendered cloud.cfg file? | |
| [19:10] <blackboxsw> besides the leading empty newlin | |
| [19:10] <minimal> it may not be the underlying cause of the issue the person is having, I'm now getting more feedback and it might be something else | |
| [19:11] <minimal> it seems they are replacing the cloud.cfg with their own version that doesn't have "system_info:" section or "distro:" subsection" so that's probably the root cause of their fallback to ubuntu | |
| [19:13] <minimal> don't know how it works for them with other distros then | |
| [19:14] <blackboxsw> for karma :) https://github.com/canonical/cloud-init/pull/1902 | |
| [19:14] -ubottu:#cloud-init- Pull 1902 in canonical/cloud-init "cloud.cfg.tmpl: drop leading newline in jinja-rendered YAML content" [Open] | |
| [19:15] <minimal> blackboxsw: I'll raise a PR to fix various cloud.cfg.tmpl warning/errors that yamllint flags | |
| [19:16] <blackboxsw> minimal: that's better. than my tongue in cheek pr 1902 :) | |
| [19:16] <blackboxsw> I'll close it | |