This repository has been archived by the owner on Apr 21, 2023. It is now read-only.
URL is not in cache #1720
Comments
|
Hi What are yours directives: How your site works? |
|
Hi, I then added pagespeed and testing since, if its needed will remove the previous ones. https://gist.github.com/uhlhosting/4ef7e41ece8ee0fc2ab8fa6049766b78 here is the uhl.site config. About uhl.site true since not caching html, alto not sure why it toke the stub url address for status from nginx into connsideration. |
|
Hi 1.- There is some files ( 2.- Why you have fastcgi directives out of a server block in uhl.site.conf? If the directives are out of a server block is the same as put it in the http block, so why don´t put it at the same level that other fastcgi directives in nginx.conf 3.- You have enbled pagespeed in the http block so pagespeed works for all the server blocks. You need pagespeed working in phpadmin or in the default server? if not, put 4.- Why you need 5.- Why you need 6.- Why do you need 7.- What means this 8.- You have 9.- You say use Redis, but in the config files Memcached is configured, not Redis. 10.- Why you need gzip enabled in .pagespeed. reosurces? 11.- If you use wordpress, maybe you need to exclude wp-admin from pagespeed: 12.- Why you use |
|
Hi, Here are the remaining details:
11.I had some dissalow rules in wordpress i first config, will introduce them again. Needed a bit more research.
Beside this, I noticed several things, my security headers seems to be ignored some of them. Also the caching time for all ressources has changed now, and is mixed some ressourced cached 5 mins, while some 30 days. |
|
Hi Maybe a reorder of these files is needed, making all directives for statistics, console and other in 1 file for each, and not spreaded in these 3. 2.- All these fastcgi are used ONLY where you have php locations, so you can put it together and have all fascgi directives only in one point and not spread it. 3.- Personally I use only 4.- I think you don´t need 5.- this if is not needed, like 9.- You can use Redis. In the docs say "experimental" but don´t is. Is safe and estable, and for me better option that memcached. 10.- You have brotli and gzip enabled in the http block. and these are applied to .pagespeed. resources too. 11.- Pagespeed may have bad consecuences in the admin area for wordpress. Better disabled for it. Same for the fastcgi cache, you don´t need the wp-login cached. 12.- You are not serving binary content by default but text/html, better change it. 13.- I see you have an other include directive i´m not aware before: Maybe I´m lossing some info with all those includes...... Maybe the panel you use have cnfigured nginx to work with a proxy, then it may work as: user -> nginx proxy -> nginx server and then some directives make sense. |
This is the proxy.conf. This it the enable-php.conf. the rewrite file is empty yet that is linked to the frontend where users can add manually rewrite rules, at the moment is empty.
Other very strange behaviour would be the page sometimes giving a blank white page, and also there is a 404 and a 501 errors hit from time to time. guess still experimenting , would need to start with less filters. I will update the newly renewed configs, once I finish them. Any idea why the timeouts of the domain when using refresh only. |
|
To debug the blank page you need to look at server logs and see what happens when the blank page come. Maybe you need to put error messages in debug level to get more info: I think that the panel had created very complicated config file structure for a simple one server config. Maybe you need some step by step process to get a whole config, I mean, make the simplified setup config as posible w/o proxy or even pagespeed, make it work clean and w/o any errors, and then intrude things like the pagespeed module. |
|
Seems that is not helping, seems no errors are recorder whatsoever: This is the most we have from access log, while error log is empty on the vhost side. |
|
Seems that even debug does not track it, there is a bug maybe? The most intresting part is that error console from nginx in the site vhost is not showing any errors. Managed to get something in error log: |
|
Hi.... Occam, a missconfiguration rather tan a bug. |
|
Changed that line: these were not previously, will check also to the admins from aapanel what they say about the config. |
|
Try this config. it is generated by this page: with the data I know for your site, maybe need some modifications. |
|
@Lofesa thanks il give it a try, anyhow that would be more complex to addapt to the deployment of nginx we have now. I have found now a recent docker deployment and im testing it out. It seems when I disable pagespeed the white pages are out, so il start slowly taking all issues one by one. Il also try to involve the aapanel admins. |
|
It seems there is some conflicts of some sort, while in the moment I enabled pagespeed, imediatelly my pages goes from 404 to 501 and back and forth to white screen.
While access console shows: And errror log: |
|
Isnt possible that nginx cache is conflicting with pagespeed? |
|
Is possibel, Ican´t use nginx cache.... I can´t get any 404 or even 501 only ERR_CONNECTION_REFUSED message. These are with the config i have send? |
|
@Lofesa i can't use that config in current environment without breaking one of them. If you refresh the page several times, you will end up seeing the page load. Like I said are several statuses that spin between them. |
|
Managed to recompile it, tweaked a bit more and removed what is not needed. Still gettings same results, I also disabled FastCGI cache. Redis cache in wp side, and left pagespeed + redis . I heard there are issues with gzip on? Should I disable gzip also, same for brotli? |
|
Latest console logs: Nginx main debug log: |
|
Is this path normal? /mnt/ram/pagespeed/shm_metadata_cache/snapshot/,2Fmnt/ram/pagespeed/metadata_cache Initially I created the /mnt/ram/pagespeed/ Is this normal for pagespeed to recreate the same path? |
|
Well....
Seems you have enabled
The problem with gzip/brotli comes when you enable gzip compression in pagespeed ( |
It seems i dont have such variable set at all. I recreated the !clean!time! and gave it rights. I have completly disabled fastcgi cache. Here is my last configs: |
|
At some point you have it
|
|
Hi, I have updated the config file and since the one you had has changed, all pagespeed variables are in the files i sent in .zip. In Safari I can notice this on console when logged in wp:
And a so called image that never seem to load properly, yet no idea what that could be, and console is not giving more. Also I noticed in chrome there is another error in console: this seems to be related with some js generated by pagespeed. What should I try next? I noticed that sometimes pages loads without pagespeed, and sometimes with it. Still getting the whitepage and on few refreshes its hittinng the pagespeed yet also without it. |
|
It seems now im warned by pagespeed and lighthouse that: I was hoping pagespeed can run with existing compression or will serve by default compressed content to avoid such warnings. |
|
Hi As far as I can see, you have excluded some files (with pagespeed Disallow) from pagespeed so these files are enterely under nginx control. This message
|
|
I have set the pagespeed HttpCacheCompressionLevel 0; . Still I hit the white blank pages, and another issue is that any link after uhl.site/testme.php will hit the blog page, and not a 404 page. So i think the last change suggested to change from |
|
It seems the errorrs are visible only when i am logged in, or maybe is browser specific will dig this. |
|
Whe I have advised about this:
This file and this line are the php location for wordpress. I use some like this in a server block: What happens in your site when pagespeed is off? (this maybe achieved with https://uhl.site/?PageSpeed=off) 2.- Are not compressed, so your config for gzip and brotli are not working, look at where you included the file with gzip and brotli config |
|
Updated fresh config file: https://gist.github.com/uhlhosting/f2ed6e353dd04c94f0997b8b31c175b2
It seems the uhl.site document is called 2 times, one time has duplicated headers, still this seem so much a bug in pagespeed or missconfiguration, while the main uhl.site is theree with normal headers, second entry it has all duplicates and a wrong chain.
There is no loadbalancer in place, this was default config, and I added the fastcgi config like here: . This remains a dilema, now you got all my latest config files. Maybe now after i reset the vhost will be easier to debug. |
|
If you disallow the domain, none of the url from the domain will be optimzed. |
|
I think also Would work, now I am not sure what would be a better or more preffered solution from the two. |
|
What when the domain is another? Like in this case:
|
|
Both of 2 are rigth, the result is what matter: url out of pagespeed. |
|
@last reply, yes there must be a server block 127.0.0.1 . for stub status and /status page from internal nginx.
The location /phpfpm_5x_status continues to all php versions up to 7.4 . Should I add unnplugged to each location, or the server should handle them all directly? Really appreciated all the time invested in replying and every feedback you provided, lifetime saver, alto I feel could write a book with all i've learned so far, still lot to learn and improve. It seems every now and then I come upon some newer configs especially for Wordpress, and I then follow up and see results. I had managed to take the pagespeed from 180 Also I was once checking 1 week ago the cache side of pagespeed, I used to have there inside 2 folders one so called v3 and the SHM folder, also I have a setting enabled, local_cache or such, you mentioned to disable it, isnt the local cache better than shm cache ? Cant they be used together? What are the debug variabled to see where issues occur and where pagespeed have troubles optimising, any good practices to find these speed gaps that keep oscilating? Here is last config version with updates on all suggestions received, some variables such as Outline* or Inline* are merely to be ready in place when Il be running tests to see what is better, to inline JS/CSS or to outline it. I have seen lots of configs where people apply both inline and outline, i know you mentioned this is wrong and should be avoided, and so I do. Version: 14: on |
|
I just notice first console error, in long time now:
Are there any pagespeed variables to keep the log folder tidy, and make sure is not overusing the space. Seems I have there entries since 24 nov. Any idea what could caused it? Never had anything to log before, the location is present and always worked. |
|
Is there a way to make the combined css's shorter especially is there even possible to change the pagespeed addition to urls, maybe into something like ps . would save lot of chars like this especially when combining urls. |
Maybe you have added more content to the page. But have:
These are diferrent caches... one for rewrited resources (those that have pagespeed in the url), one for metadata about resources (what url belongs to a webp version of a jpg url ando so..). SHM is a in memory cache, but you have enabled to write it to disk every 5min (ShmMetadataCacheCheckpointIntervalSec) to survive between restarts. Local cache (pagespeed EnableFilters local_storage_cache;) is for others purposes. Read https://www.modpagespeed.com/doc/filter-local-storage-cache.
https://uhl.site/?PageSpeedFilters=+debug
Inlining maybe wrong when http2 is in place.
Maybe you have deleted a file in use... There is no directives to clean up the statistics files.
No and no. Maybe you can combine less files with |
|
Maybe you have added more content to the page. But have: What would be the default, is there a way to see all default variables including those I did not setup? |
|
The place to see it is the doc web. whit How the deadlines work? If the resource takes more than the configured time, pagespeed serve a unoptimized url AND in backgruond it optimize the resource so the next time the resource is requested it get optimized. You can wait to the resource optimization, but these time you wait is added to the request completion. |
This si always my starting console log, any idea what is causing all the duplicated entries in same time? Is this normal behaviour? And from where the part? |
|
In how many servers blocks have you enabled the shared memory path? or in how many servers block have you an include file that contains the shared memory path?
Have you set a shared memory path in a server with |
Not that I know of, no havet done this. |
|
As per this commit maybe you have a mix of server block/location where cache purge is on/off. You need to trace where pagespeed is set to unplugged and don´t set shared memory, and where the |
|
pagespeed EnableCachePurge is set to on/off. only in one location. I havent changed the structure of the files sincee last i sent them. |
|
In the last set of files you have posted, the directive is in main.conf. Where have you included this file? |
|
Hi @Lofesa I was a bit off server work, now back into it. [Tue, 12 Jan 2021 08:40:05 GMT] [Info] [718939] Trying to serve rewritten resource in-place: https://uhl.site/ I am only getting this mostly nowadays in console, Isnt a way to find out why isnt the whole site indexed? Or whats causing this info. Can it be ignored? If thats the case why the repetition. |
|
That´s normal log. |
|
Need to check documentation for log level , for now havent seen anything to change this within my settings. |
|
I am keep receiving in nginx console these: 2021/01/16 13:58:25 [warn] 445981#445981: pagespeed: cannot set gzip, command not found in /www/server/nginx/conf/nginx.conf:18 Now I havent altered the nginx.conf . |
|
What pagespeed/nginx version have you? |
|
Pagespeed latest stable, and nginx 1.19.6 . I did restarted yet it will reappear the message. |
|
Line 18 is |
|
No that line is in another file thats called in nginx.conf. In nginx conf i have normal gzip on; and brotli on; |
|
@Lofesa not sure why this is now happening it seems somehow pagespeed ignored my variables: WordPress-specific configurations, change according to your domain and installation pathpagespeed Disallow /.php*; I specifically specified not to change those files, when I try to use the oxygen builder editor I get this: |
|
These are just regex expressions: Maybe |
|
@Lofesa I am also testing CSP with pagespeed and it seems that there is some external ongoing connection to www.pagespeed-mod.com ! ? Any idea what would pagespeed need to connect to external sources? |
|
I have tested https://uhl.site in chrome browser and in webpagetest and can´t see any outgoing connection to www.pagespeed-com.com, only uhl.site and stats.uhl.site But you have some mistakes: |
|
@Lofesa I believe this is the way Pagespeed inlines the css if its in header... This is purely done by pagespeed. Maybe it can be more investigated and a bug raised? 2.-You have 1 url (https://uhl.site) with 2 cannonicals (https://uhl.site and https://uhl.site/de). These is not the way to have multilingual pages. You need to explore the hreflang html labels. Can you elaborate a bit on this one? I can invite the developer of the plugin since this is very early alpha stage plugin for oxygen. @sbouey maybe you can check this? |
|
1.- You have this html code: and have enabled: cj Move Css Above Scripts So css are inlined before the 1st script tag. Put all script tag at the bottom of the head section. 2.- How canonnical work, you can have 2 or more url with the same canonical, so bots (like Google bot) can index, and show in search, only the canonnical, w/o penalize with duplicate contents. For example amp pages. An amp page has the same content as the normal page, w/o canonical, google can penalize these pages (amp and normal) with duplicate content. For having sites with different languages you must tell to search engines that a page have related pages for diferent languages, this can be achieved by diferent paths: htl code, http headers or sitemaps. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Hi,
After deploying pagespeed on my server, tested several configs and options, yet lately these info messages are filling the console, was wondering if anyone knows how to handle these:
Attached the console messages in a gist.
https://gist.github.com/uhlhosting/c6bf91d643b8fc0d86f6459eddb80e4c
Running on CentOS 8, with nginx 1.19.4.
The text was updated successfully, but these errors were encountered: