aciba[m] | Hello DP84: cloud-init looks for config modules in the current dir and `<installation-path>/cloudinit/config`. The current dir is useful in the case where `cloud-init init|modules` is manually run but, when run as systemd unit, the current dir is tipically / | 07:36 |
---|---|---|
aciba[m] | Code exercising this: https://github.com/canonical/cloud-init/blob/f755bad5d3c683a7102acdf053c4121a5d568be8/cloudinit/config/modules.py#L195 | 07:36 |
Akila | hi, I need help to know: how to provide a seed-url of a webserver or public cloud that needs authentication...like auth-token or bearer token...how and where to specify it | 19:08 |
Akila | in datasource | 19:08 |
minimal | are you the person who raised an Issue about this recently? | 19:24 |
Akila | no...first time raising this query | 19:25 |
minimal | so you're using a public cloud but not using their specific DataSource? | 19:28 |
minimal | holmanb: #4271 is not a bug, see the note I added | 19:28 |
Akila | does thi mean that the public cloud should enable/support cloud-init, thereby exposing a datasource rule to use.. | 19:30 |
minimal | Akila: which cloud is this? | 19:33 |
Akila | trying to see if we could get this supported in hpe greenlake | 19:34 |
minimal | Greenlake appears to support cloud-init but no mention of which DataSource is used | 19:36 |
minimal | https://support.hpe.com/hpesc/public/docDisplay?docId=a00092451en_us&docLocale=en_US&page=GUID-BEEE4F70-7A4F-4247-8DBC-91FF1502F970.html | 19:37 |
minimal | Akila: a seed url is only used by the NoCloud datasource in cloud-init | 19:37 |
minimal | typically for a cloud provider you use a different (cloud-specific) datasource | 19:38 |
Akila | based on what I read I see 2 ways: 1. datasource with seed-url : having url of the service/webserver that can provide me the config..But how to pass token if that is looking for authentication.... | 19:40 |
Akila | 2nd method is: each cloud provider has given a config to use: like in datasource: [EC2]: then URL and some additional config | 19:40 |
Akila | for 2nd option, then I need to check with greenlake cloud team on datasource to use.... | 19:41 |
Akila | any idea for 1st option to specify a token if I use as Nocloud | 19:41 |
minimal | Akila: typically you should use the relevant datasource for that cloud in order to get configuration such as network config | 19:43 |
minimal | as the cloud provider typically "controls" network configuration | 19:43 |
Akila | Usecase is bit different ..i'm trying to use a configuration( supplied by cloud) on a on-prem server... | 19:48 |
Akila | via cloud-init | 19:48 |
minimal | "configuration (supplied by cloud)"? what configuration | 19:48 |
Akila | like scripts to apply, user-data and meta-data will come from cloud (user enters these values in cloud interface) | 19:51 |
minimal | regarding authentication for seed URLs, this is not supported. Where would you expect any authentication info to come from? | 19:52 |
minimal | Akila: I'm confused, you just said it is on-prem, but now you're referring to "cloud" and "cloud interface" | 19:53 |
minimal | so is this Greenlake running on-premises as a "private cloud"? | 19:54 |
Akila | ok..the use case is: i'm a cloud user..using cloud service I manage my on-prem servers..would like to install OS on those servers and configure as well...this configuration will be supplied by user via cloud interface | 19:54 |
minimal | and where does network configuration information for servers come from? | 19:56 |
minimal | or do they just DHCP on (all) network interfaces? | 19:56 |
Akila | yes..to start with let us take dhcp is the setting | 19:58 |
Akila | when server comes up, it is dhcp - gets IP | 19:59 |
minimal | so if you want to use seed url then you need to use NoCloud datasource and provide the url somehow (pass it from bootloader in cmdline or hardcode it into a file in /etc/cloud/) | 20:00 |
minimal | then cloud-init will fetch "meta-data", "user-data" etc by appending those to the "base" url specified | 20:01 |
minimal | however as I already said there is no authentication support for this | 20:01 |
minimal | also you may find some problems with using seed URLs | 20:01 |
holmanb | minimal: just saw your comment on that bug, see my response | 20:03 |
minimal | holmanb: ah, I thought it was a copy of the /etc/cloud/ file | 20:04 |
minimal | I was confused as there was mention of passing the seed stuff on the cmdline | 20:05 |
holmanb | minimal: all good - I think the fix is a one character change in ds-identify, but I'm working on an integration test to cover this use case now. | 20:05 |
minimal | holmanb: I still need to get around to raising an issue about seed-url handling in general - just need to find the time to do it | 20:06 |
Akila | ok | 20:11 |
holmanb | minimal: gotcha | 20:31 |
holmanb | minimal: got a fix up https://github.com/canonical/cloud-init/pull/4273 | 20:32 |
-ubottu:#cloud-init- Pull 4273 in canonical/cloud-init "Kernel Commandline: Fix 4271" [Open] | 20:32 | |
minimal | ah, a fix of a fix, the best type ;-) | 20:32 |
holmanb | and a fix of my own fix, so no one to point fingers at but myself ;) | 20:33 |
minimal | I'm not one to point fingers, especially as I'm trying to fix the PR I just opened lol | 20:34 |
holmanb | heh | 20:34 |
holmanb | and that reminds me, we can un-document nocloud-net now that nocloud can automatically detect mode | 20:34 |
=== jrm2 is now known as jrm | ||
minimal | all those test fails were due to just a single missing quote, lol | 21:39 |
meena | minimal: don't you love programming | 22:00 |
minimal | its a love-hate relationship lol | 22:01 |
minimal | and now I'm trying to figure out why a single integration test failed :-( | 22:01 |
minimal | meena: I remember programming on batch systems in the 80s, you submitted your compile job and came back the next day to see the result - you soon learning to carefully check for typos after a few days of failed compiles ;-) | 22:30 |
meena | minimal: i was born in the 80s. but I was also born in Bosnia, so it's kind of like being born in the late middle ages | 22:37 |
meena | I didn't programme, or even use computers until the turn of the millennium | 22:37 |
minimal | it was more fun back then lol | 22:47 |
blackboxsw | the 70s were way better. given your start of programming. you may have missed out on programming in logo and the turtle | 22:48 |
blackboxsw | crucial moments in early education and development :). | 22:48 |
meena | blackboxsw: i started programming in C, got some pascal on the side, then we had C++ and pic assembler, then Java and PL/I with some REXX and S/360 assembler. after that, i mostly got by with perl, Ruby and php for a long time | 22:56 |
meena | lots of 70s tech there… | 22:56 |
meena | C, REXX, PL/I, S/360 are all 70s tech. honorary mention for php, a razorblade thin layer on top of C, and Ruby, a Smalltalk | 23:03 |
minimal | I did BCPL, Pascal, COBOL, Fortran, Lisp, Prolog, Basic, 6502 assembler | 23:30 |
minimal | and bits of Modula-2, C, 68000 assembler, Poplog, and Z80 assembler | 23:31 |
minimal | dBase, even did some FI/PHP (the first version of PHP), Netscape LiveWire (server-side Javascript long before NodeJs) | 23:32 |
minimal | oh, Forth as well, not sure if I did any Logo | 23:33 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!