How can I fix this problem?

@RoldanLT, I've been saying to disable ngx_pagespeed long time ago... it is useless on a one site setup.
depends on the site and style deployed

no problems for past 33+ days on my forums https://community.centminmod.com after fine tuning my ngx_pagespeed filters. Using Centmin Mod based Nginx 1.7.2, PHP 5.5.14, MariaDB 5.5.38 + CSF Firewall + Google SPDY SSL + ngx_pagespeed :)

primed ngx_pagespeed gtmetrix result = one site setup :D

note: both twitter and google login integration also enabled

Dallas, Cable 5Mbps test

gtmetrix_030714.webp

pagespeed_030714.webp
 
Last edited:
@eva2000, we can talk this for hours, I will never install pagespeed on any of my servers. :)
Results with Nginx cache disabled and obviously no pagespeed:
http://gtmetrix.com/reports/www.axivo.com/XRd1scqj
http://www.webpagetest.org/result/140704_T8_1KP/

axivo-firstbyte.webp

axivo.webp

The 'low' score is not related to server setup but external files. Example for Google webmasters page, the bible of optimizations:
http://gtmetrix.com/reports/www.google.com/Oy98scfD

google-webmasters.webp

People are just getting paranoia on miliseconds, for myself and any conscious sysadmin is more important to conserve the precious server resources. Try pagespeed on a huge site and let me know how it ruins your multiple nodes performance.

What lowered my score... do I really care? :giggle:
Leverage browser caching of static assets: 77/100
FAILED - (No max-age or expires) - https://platform.twitter.com/widgets.js
FAILED - (No max-age or expires) - https://accounts.google.com/ServiceLogin?service=lso&continue=https://accounts.google.com/o/oauth2/auth?zt=ChRwOUV3TTdDU3V5V0F0MFZYTEpoNRIfa2d4MWNicGw2bThXM29FQmQ4RE90TkNmclBfOWpRSQ%E2%88%99APsBz4gAAAAAU7YHGSA3q-mlyJeHwTLg_1r3mXj4Wnkv&from_login=1&hl=en-US&as=-aaa5bc8139c2e3c<mpl=popup&shdf=CrQCCxIRdGhpcmRQYXJ0eUxvZ29VcmwayQEvL2ltYWdlcy1sc28tb3BlbnNvY2lhbC5nb29nbGV1c2VyY29udGVudC5jb20vZ2FkZ2V0cy9wcm94eT91cmw9aHR0cHM6Ly93d3cuYXhpdm8uY29tL3N0eWxlcy9kZWZhdWx0L3hlbmZvcm8vbG9nby5nby5wbmcmY29udGFpbmVyPWxzbyZnYWRnZXQ9YSZyZXdyaXRlTWltZT1pbWFnZS8qJnJlc2l6ZV9oPTEyMCZyZXNpemVfdz0xMjAmbm9fZXhwYW5kPTEMCxIVdGhpcmRQYXJ0eURpc3BsYXlOYW1lGgVBWElWTwwLEgZkb21haW4aBUFYSVZPDAsSFXRoaXJkUGFydHlEaXNwbGF5VHlwZRoHREVGQVVMVAwSA2xzbyIUXHulwbiurn9KqpkHUzyTz8v1sFYoATIUaoLvNmiKdfyt4-VLmiCFej2DV-Q&hl=en-US&passive=true&go=true&sarp=1&scc=1
FAILED - (No max-age or expires) - https://plus.google.com/js/client:plusone.js
FAILED - (No max-age or expires) - https://apis.google.com/js/api.js
FAILED - (15.0 minutes) - https://cdn.api.twitter.com/1/urls/count.json?url=https://www.axivo.com/forums/&callback=twttr.receiveCount
FAILED - (20.0 minutes) - https://connect.facebook.net/en_US/all.js
WARNING - (60.0 minutes) - https://oauth.googleusercontent.com/gadgets/js/core:rpc:shindig.random:shindig.sha1.js?c=2
WARNING - (12.0 hours) - https://ssl.google-analytics.com/ga.js
WARNING - (24.0 hours) - https://accounts.google.com/o/oauth2/postmessageRelay?parent=https://www.axivo.com
 
Last edited:
haha yeah know.. personal preference, i have some with and some without ngx_pagespeed all depends on the site and preferences :)
 
@eva2000, indeed... I prefer to have a site that has a very low firstbyte with keepalives properly set, instead of optimizing JS just to kill my servers. :D
For those who do not know how pagespeed works: it doesn't manage to rewrite everything the first time. While the processed code is passing through, it needs to get the referenced resources, rewrite them and put them together. That process can take a while and when it takes more than a few ms, it just sends on the html unmodified or partially modified. If you cache the output of pagespeed immediately you'd be caching only a partly rewritten page... What do you you think this does to your server resources?
 
Last edited:
Well ngx_pagespeed won't cache partially rewritten/optimised pages natively so that isn't a concern. And well there's no free lunch so yes faster page speeds do use more resources but it isn't as bad with ngx_pagespeed as it is for mod_pagespeed for Apache. I definitely get much faster initial render/visual times with pagespeed enabled than disabled according to my GA stats across all geographical visitor regions.

ga_pagespeed_040714.webp
 
  • Like
Reactions: rdn
What are the numbers with pagespeed disabled? I guarantee you the difference is very minimal.
I have no idea how you got those stats in GA... Can you please teach me where to get them?
 
yeah they're under site speed in GA just you use custom Google Analytic dashboards to tailor the stats gathered https://support.google.com/analytics/answer/1068216?hl=en and add your own widgets https://support.google.com/analytics/answer/1068218?hl=en&ref_topic=1068215 :)

The difference is ~0.25 to ~0.45 seconds faster in pagespeed in terms of initial render/visual time which for my forums is a difference between 0.75s to 1.25s perceived load time (visual render time) vs 1.00s to 1.50s on average. That's decent for a VPS server, with dedicated servers, I usually see another 0.15s to 0.25s improvement depending on the geographic location.
 
@eva2000, on my tests I was getting like 0.10 seconds max gains on an overall page load of 2 seconds... that in my books is nothing. I'll rather make the user wait a tenth of a second and avoid the huge resources usage.
 
yeah as i said depends on site i.e. style used and page elements etc

hence, why i said some sites i use ngx_pagespeed and some I don't :)
 
just don't get addicted to GA dashboards... once you start you maybe not be able to stop as it's such a powerful and useful feature in GA :)
Yes, I'm completely addicted to it but I think I'm getting to the end of the road with it. What's great is, I created one that I'll be able to export as PDF to send to potential advertisers. That way they get everything they need:
Ages
Gender
Uniques
Pageviews
Interests
Breakdown of devices (desktop vs mobile vs tablet)
And maybe one or 2 other things

I think sending a PDF like that would be much more powerful than to just simply state the stats. gA makes it super easy once you figure out how to use the dashboards.
 
Back
Top Bottom