I’m noticing on ApisCP, and especially via the access logs, that 90% of the bandwidth is coming from the server itself, via mod_pagespeed. It shows the source IP to be the same IP as my ApisCP server.
Does local traffic via pagespeed count as bandwidth? If so, is there a way to omit that?
Alternatively, are there any other ways to get a more detailed breakdown of traffic so I can get an idea of where its coming from and to (other than analyzing the access_log, which I currently do)?
Pagespeed will make 1 request per asset per domain until TTL expiry. If you have multiple domains masquerading as a single domain, then mapping origin will consolidate requests. Additionally, if cache control headers aren’t set, TTL defaults to 5m.
<FilesMatch "\.(gif|jpg|png|js|css|eot|svg|woff2|json|woff)$">
FileETag none
ExpiresActive On
ExpiresDefault A864000
</FilesMatch>
This will limit self-checking common static assets to every 10 days.
Second, I see how this results in erroneous accounting. It’s been patched in edge with an updated Apache release to filter internal requests.
yum clean all
yum update -y httpd
upcp
EditDomain --reconfig --all