Remove all references to settings.BASE_PATH
#189
Conversation
| @@ -156,6 +152,9 @@ EXEMPT_VIEW_PERMISSIONS = [ | |||
| # 'ipam.prefix', | |||
| ] | |||
|
|
|||
| # If hosting Nautobot in a subdirectory, you must set this value to match the base URL prefix configured in your HTTP server (e.g. `/nautobot`). When not set, URLs will default to being prefixed by `/`. | |||
There was a problem hiding this comment.
Still learning how this works. If the HTTP server is setting SCRIPT_NAME correctly, then FORCE_SCRIPT_NAME is not required, correct? This would only be needed if SCRIPT_NAME is incorrect?
There was a problem hiding this comment.
MOST of the application works correctly without deliberately setting FORCE_SCRIPT_NAME, but there are a few edge cases mostly related to not being able to leverage django.urls.reverse() consistently due to circular imports caused by having to end up loading nautobot.core.urls where nearly every single model in the project is loaded to facilitate nautobot.core.views.HomeView and SearchView.
So for now, to allow for places where APISelect widgets are being used, namely the Device.position dynamic lookup, this is the only way to assert that we can reliably make the app receive the correct SCRIPT_NAME. If we can refactor enough to be able to use reverse() everywhere we need to specify URLs, we can stop hard-coding API URLs in the code and can simplify deployment even further.
|
Fixes: #166 |
|
|
||
| If the location of SSL certificates had to be changed in `Obtain an SSL Certificate` step, then the location will need to be changed in the `apache.conf` file as well. | ||
| ```apache |
There was a problem hiding this comment.
I think you will need to add apache to the hljs_languages: list in mkdocs.yml for this to get highlighted appropriately, though I'd love to be wrong on this point.
There was a problem hiding this comment.
It would also be great to add nginx highlighting to the NGINX config earlier in this file, and enable it in the mkdocs.yml as well.
There was a problem hiding this comment.
Tried it and it didn't seem to do anything. 🤷
Yeah I wasn't touching that doc just yet. It's still on the todo list and I figured I'd just do that then. |
This now relies on the HTTP CGI variable `SCRIPT_NAME` to be passed in upstream from the HTTP server (e.g. NGINX) paired with matching uWSGI configuration. This also refactors social auth URL to never rely upon `BASE_PATH` and instead replace hard-coding `LOGIN_URL` w/ a template condition on `settings.SOCIAL_AUTH_ENABLED`. Further, this means that the social auth URL can naturally inherit the derived base path sent from the HTTP server without having to hard-code it. - Correct references to hard-coded URLs and instead let them reverse properly - Update uwsgi settings to include optional setting for subdir hosting - The default values of `STATIC_URL` and `MEDIA_URL` are now relative URL paths to allow for them to inherit the derived base path. Under normal circumstatnces this is equivalent to them being absolute. - In the event that a HTTP server sets the base path with `SCRIPT_NAME`, (e.g. `/nautobot`) then these automatically get prefixed (e.g. `/nautobot/static/`). If a user requires custom static/media paths, they may still customize these values for their setup. - Generic view for `nautobot.core.views.HomeView` updated to `TemplateView` to help in template debugging.
- Make extras forms call api_url w/ a route name vs. URL path
This also splits core API URLs into `nautobot.core.api.urls` and `include` those URLs in `nautobot.core.urls` to reduce the amount of circular dependencies going on in `nautobot.core`
- Update production configs to remind use of `FORCE_SCRIPT_NAME` to match that of HTTP config - Update config init template to include `FORCE_SCRIPT_NAME` - Fix URL join in `widgets.py` - Added a basic unit test for `FORCE_SCRIPT_NAME`
Update docs to reference subdir prefix using trailing slash for consistency, since this is what Django emits it as.
- Fold `apache.conf` into install docs for http-server - Update `FORCE_SCRIPT_NAME` test to use `try...finally`
jathanism
force-pushed
the
jathanism-replace-base_path
branch
from
03e16cf
to
f0127d1
Compare
|
Just rebased against upstream. |
Fixes: #147
This now relies on the HTTP CGI environment variable
SCRIPT_NAMEto be passed in upstream from the HTTP server (e.g. NGINX) paired with matching uWSGI configuration and thatFORCE_SCRIPT_NAMEbe set innautobot_config.pyto match the configured path.This also refactors the social auth URL to never rely upon
BASE_PATHand instead replace hard-codingLOGIN_URLw/ a template condition onsettings.SOCIAL_AUTH_ENABLED. Further, this means that the social auth URL can naturally inherit the derived base path sent from the HTTP server without having to hard-code it.The default values of
STATIC_URLandMEDIA_URLare now relative URL paths to allow for them to inherit the derived base path. Under normal circumstances this is equivalent to them being absolute.In the event that a HTTP server sets the base path with the HTTP environment variable
SCRIPT_NAME, (e.g./nautobot) then ALL URLs automatically get prefixed (e.g./nautobot/static/). If a user requires custom static/media paths, they may still customize these values for their setup.Other updates:
nautobot.core.api.urlsandincludethose URLs innautobot.core.urlsto reduce the amount of circular dependencies going on innautobot.coreLOGIN_URLandLOGIN_REDIRECT_URLare now using URL route names and not hard-coded pathsuwsgi_param) config directives, and to include optional setting for subdir hostingsocket(uWSGI protocol) vs.http-socketwhen behind NGINX anduwsgi_paramto configure NGINXnautobot.core.views.HomeViewupdated toTemplateViewto help in template debugging.social_djangoso that various auth backend variables can be used in templates.social-auth-app-djangoas a hard dependency to simplify dynamism in the SSO login URL.FORCE_SCRIPT_NAMEto match that of HTTP configFORCE_SCRIPT_NAMEFORCE_SCRIPT_NAMEcontrib/directory once and for all and all references tocontrib/(except in upgrading doc, which will come later)