Closed bblakely-anl closed 2 months ago
This change should not be needed anymore - the problem seems to have been fixed in the new versions of the toolkit. Can you confirm that it now works properly without this fix/workaround for you as well?
Thanks for the suggestion @bblakely-anl!
We've taken a different approach in the mean time and downgraded all internal traffic to IPv4 by using 127.0.0.1
instead of localhost
addresses. Closing.
Description
For me, there was an issue where sharelatex/nginx were trying to use 127.0.0.1 to access the proxied services, but some were only listening on ::1, and one seemed to be listening on ::1 but redirect to 127.0.0.1. To make things consistent, I disabled IPv6 in that container to ensure they all use 127.0.0.1. An alternative solution would be to make all config references consistent (127.0.0.1 vs "localhost", which may resolve to ::1), but this seems simpler and IPv6 is unlikely to be needed for the internal Docker network in this case.
Related issues / Pull Requests
None
Contributor Agreement