I have a page was requesting 'assets/ember/scalus-services.js'. These assets did not exist, since the assets are built by Ember CLI with an MD5 fingerprint
i.e - 'ember/assets/scalus-services-000aaa111222eee333333111222eee33.js'
Hence I received:
Failed to load resource: the server responded with a status of 404 (Not Found)
When using asset helpers like asset_path, this path is resolved one of three ways
-
It looks it up in manifest-.json (loaded into memory as Rails.application.assets) under the 'assets' object. the assets object is basically the path you would pass into a helper like 'ember/scalus-services.js' is the key and the fingerprinted file name is the value. This value includes the cloudfront prepended path (which is set in the Brocfile.js) and the rails asset path ('assets/')
-
If the file is not found in the manifest file, it will search the asset paths in rails for the file
-
If it isn't found in the rails asset paths, the path is served as is
The problem though is that all servers had a manifest file with the correct entries in 'public/assets'. Yet it served the path up as is.
This leads me to believe that this is a bug with sprockets or perhaps the deploy process.
HOWEVER: This bug didn't happen immediately after deployment. It started 12 hours after I deployed.
Aucun commentaire:
Enregistrer un commentaire