Closed callahad closed 11 years ago
Imma ganna guess that this is on default with the Persona stack that is used. But worth roping in @gene1wood
Looks like this is happening for most of our documents, but not the well-known file or heartbeat:
$ curl -s -I -H 'Accept-Encoding: gzip,deflate' https://google.login.anosrep.org/static/js/provision.js | grep Content-Encoding
Content-Encoding: gzip
$ curl -s -I -H 'Accept-Encoding: gzip,deflate' 'https://google.login.anosrep.org/static/css/style.css' | grep Content-Encoding
Content-Encoding: gzip
$ curl -s -I -H 'Accept-Encoding: gzip,deflate' https://google.login.anosrep.org/__heartbeat__ | grep Content-Encoding
$ curl -s -I -H 'Accept-Encoding: gzip,deflate' https://google.login.anosrep.org/.well-known/browserid | grep Content-Encoding
Those are tiny. Many there's some kind of size threshold before gzip kicks in?
Hm, nope. auth_failure.js is about the same size as the well-known file, but does get gzipped.
@gene1wood what's up with gzip in stage? :)
So here's your specific nginx config : https://github.com/mozilla/identity-ops/blob/master/chef/cookbooks/identity-bridge-gmail/files/default/etc/nginx/conf.d/idbridgegmail.conf
And here's your general nginx config which this is included from : https://github.com/mozilla/identity-ops/blob/master/chef/cookbooks/persona-common/files/default/etc/nginx/nginx.conf
I'd expect that the compression settings will be found in there. If so free free to send a pull request
Yea, as I expected, it's related to content-type. __heartbeat__
returns plain text, and /.well-known/browserid
returns application/json
. That config file says it gzips for html, css, and javascript.
Heartbeat doesn't matter, since it's 2 separate characters, it can't possibly benefit from gzip.
Hooray, closing as working :)
This should probably be done at the nginx layer, right?