| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
Previously we supplied the authorization information via the 'headers' parameter. This works fine, except in some cases when the request is retried.
There was a problem hiding this comment.
Thanks, read the associated context and this makes sense. Re-ran the autobuild test that is known to time out.
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Previously we supplied the authorization information via the 'headers' parameter and left the auth parameter undefined. This works fine, except in some cases when the request is retried.
We observed difference in behaviour, when using undefined for the auth parameter we got a 504 error on the first attempt and a 401 error on the second attempt. When the auth parameter was set we observed a sequences of 504 errors until the number of retries was exceeded. I'm not really sure where the Authorization header got lost, but I suspect the HTTP client drops it when following redirects and the retry-helper doesn't restore it before trying again.
In case someone else has similar problems: the underlying cause of the 504 errors were due to proxy settings. When downloading a release artifact from a GitHub Enterprise server (say github.company.com) you get redirected to media.github.company.com. The no_proxy variable did not contain the media subdomain which caused a gateway timeout. Explicitly adding the subdomain to the no_proxy variable solved the issue. Alternatively the following pull request actions/toolkit#1223 would have solved the issue too, because it updates the no_proxy handling to automatically include subdomains.
Merge / deployment checklist