-
Notifications
You must be signed in to change notification settings - Fork 32
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
update_engine failure to update when passing through a proxy #1326
Comments
I'm no rust expert, and a library may blind-sight me, but I'm not seeing anything in there to handle proxy cases. Am I wrong ? |
Correct, thanks for the bug report and testing Alpha! This case slipped through in the migration. Indeed we should add code to support the |
Thanks a lot for the confirmation @pothos , really appreciate the quickness ! May I suggest to look into supporting more than just the "ALL_PROXY" ? curl and wget have evolved to support many other variables (e.g. SOCKS_PROXY, NO_PROXY, HTTPS_PROXY, etc.). As curl was used previously, I believe the customer base would expect the behaviour to be maintained, so preferable to get curl's quirks than other binaries. I'm just saying this to avoid other customers to have problem as well but having a different implementation then us. By the way, we did test most of those variables in trying to fix our issue. :) |
By the way, we're not testing "alpha". We're on "stable". I don't think it changes anything, but I wanted the fact to be straight. |
Ah, so for Stable we use curl in the postinst script action for downloading, not sure why this doesn't work then - maybe the env var is not passed through. |
The ue-rs download_sysext helper also supports the needed proxy env vars when invoked directly. |
Filed a PR and will test when the image build is ready: flatcar/update_engine#37 flatcar/scripts#1599 |
The fix is merged and will land in the next round of releases for all channels |
FYI, the fix is required in the OS that does the update. If you don't have connectivity without the proxy, you can do an airgapped update with |
Report from the other thread: |
We are thinking about adding workaround for the two issues of proxy settings and custom payload URLs to make it work with any old client. Will post when done. |
Workaround is rolling out for Stable, Beta, Alpha - hope all is good now. |
We were able to see the nodes update. Some will update tomorrow, so I'll
close the ticket when we've done them all.
Regards,
Éric Lafontaine
…On Wed, Feb 14, 2024 at 5:16 AM Kai Lüke ***@***.***> wrote:
Workaround is rolling out for Stable, Beta, Alpha - hope all is good now.
—
Reply to this email directly, view it on GitHub
<#1326 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAXCYVDTQ4B43NWEAHRSBITYTSFJPAVCNFSM6AAAAABCCRKZU2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNBTGQ2TMMJWHA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
Description
The update process used to work through proxies with environment variable "ALL_PROXY". Today, we've realized that the update process have failed for all machines that are being the proxy.
This is what we used to inject into the update-engine.service through a dropins;
Impact
Machines behind a proxy aren't updating anymore.
Environment and steps to reproduce
Expected behavior
Get an updated machine on latest version of flatcar.
Additional information
We've tried to understand what happened, and we believe it's due to some change of the curl library being removed and another fetcher being used in the flatcar-postinst script. The reason we believe so is that we were able to test the download which seems to be successful, but something fail after. We believe it might be linked to the following change ;
flatcar/update_engine@43bb354
We'll keep on posting if we ever find the issue for sure. Please let us know if you require anything else from us.
The text was updated successfully, but these errors were encountered: