Closed balazshasprai closed 1 year ago
Thank you! Sorry about the CI woes. It's been pretty flaky lately.
Yeah no worries. I thought it might override the already existing proxy env vars on the runner/vagrant hosts, but according to this (printing out the ansible_facts), they're not set. So that shouldn't(?) be an issue?
Now I'm thinking it might not like the default('')
, checks are currently running for a default({})
instead.
If that works, I'll tidy up the environment: in the playbooks a bit and add a new commit to this
Seems like it's fine now, if vagrant decides to cooperate. Please retry the checks, reran them on the fork, they're all nice and green
Proposed Changes
Added the option to install from behind an http proxy.
If configured (false by default), it puts your
http_proxy
+https_proxy
+no_proxy
env vars into ak3s(-agent).service.d/http_proxy.conf
file, before anything gets installed. This way k3s can access the internet for images etc through the proxy.Also added env vars to the playbooks, defaulted to '', so ansible can still download the binaries and manifests even if you're behind a proxy.
In the reset main.yml I thought I'd add a conditional task. So it doesn't try to remove the http_proxy files and errors out when you don't have them enabled/installed.
(I've ran and re-ran the checks a bunch of times on my fork, not sure what's up with the hosted runners. The majority of the time it failed with the gurumeditation state error and when it did go past that, it couldn't resolve DNS. I've ran this both with http_proxy enabled and disabled, installed fine in both environments.)
Checklist
site.yml
playbookreset.yml
playbook