• Sign up or login, and you'll have full access to opportunities of forum.

Problem With Images

Go to CruxDreams.com
Yes. That should be easy to change. I shall bring it to Eul's attention.
Thanks for those helpful suggestions, I'm working on an improved/ updated version. I've made a few initial changes, but need colleagues' agreement on the more technical and moderator-specific aspects.
Thanks much! :) FYI, my suggestion with CDN went nowhere unfortunately, as I couldn't make the conversation going with ImageMaker on that matter. I'm glad that I've done at least something positive to help mitigating the problem in a different way, thanks to your efforts.
 
Last edited:
Can anyone help, please?

I've heard from a member who is having problems seeing full-size, externally-hosted images, such as the big ones on News from Margot - he just sees 'dud' icons where the pics should be. This seems to have begun after the recent software upgrade.

There are also reports of some other external-hosted images bringing up error messages when members click on them.

Has anyone here experienced any such issues, or heard of them from other members?
(Or indeed, of any other transmission problems)
 
Can anyone help, please?

I've heard from a member who is having problems seeing full-size, externally-hosted images, such as the big ones on News from Margot - he just sees 'dud' icons where the pics should be. This seems to have begun after the recent software upgrade.

There are also reports of some other external-hosted images bringing up error messages when members click on them.

Has anyone here experienced any such issues, or heard of them from other members?
(Or indeed, of any other transmission problems)
Any videos I host from Imgur.com lack audio. I'm not sure it's related to the upgrade. It could just be coincidental.
 
The page appear like this: View attachment 905370
I can reproduce this problem in Chrome on mobile. In Firefox on Mobile which is what I normally use, it doesn't happen.

Interestingly, if I press my finger down on the broken image icon, after a while I'll get a popup menu that allows me to "open image in a new tab" among other things ... and that works.

Since the images load normally in Firefox I'd assume this may be an issue with Chrome only, perhaps a recent update, and not CF as such.

I've verified just now that the problem doesn't show up on Chrome Desktop at least for me.
 
The page appear like this: View attachment 905370
Does this happen on other threads with embedded full size images too, or only on this one, Margot's?

As a control, use this one for comparison: https://www.cruxforums.com/xf/threads/the-soppy-sentimental-photostory-thread.8280/
Do the full-size images show or are they also just broken icons?

The reason I ask:

On Margot's thread, images are embedded from http://redfeline.com/wp-content/uploads/...
the http:// tells us it's an insecure connection so Chrome mobile may (in the interest of protecting your phone) refuse displaying 'mixed content' by default. It will also give a 'page insecure' warning in the address bar.

While in the other thread the images come from https://i.imgur.com and so it's secure.
 
The last days, it happened to me that, when clicking on a thumbnail to enlarge, a new page opens, asking me to login, and only then I can view it. Meanwhile, the original page I was using, remains open.
 
Can anyone give me a link to the post which can reproduce the problem? (Never mind it, just found it... had to search the forum for "bastinado" :D)

I've heard now of some members getting 'error' messages - indeed I have a couple of times myself when navigating through the site, but only momentarily.

If the error message contains something like "Nginx", "Mysql", "Bad Gateway", or any 3 digit number starting with 5 (e.g. 500, 502, 503), it's from the server itself. Sometimes Cloudflare can detect this and show it in a bit more user-friendlier manner like this:

EhGbL.png

(Image linked from stackoverflow.com)

If you see 5xx number in the title or the red icon on the "Host" part, it means that the CF's server or its network has some problem.

I also experienced a few occasional problems like this in the recent past, including an error message about Mysql connection. These can be temporary problems but if they persist, we need to look deeper to find a cause. It's a different kind of problem from the image issue we used to have and a CDN doesn't help nor aggravate it, generally speaking.
 
Last edited:
Does this happen on other threads with embedded full size images too, or only on this one, Margot's?

As a control, use this one for comparison: https://www.cruxforums.com/xf/threads/the-soppy-sentimental-photostory-thread.8280/
Do the full-size images show or are they also just broken icons?

The reason I ask:

On Margot's thread, images are embedded from http://redfeline.com/wp-content/uploads/...
the http:// tells us it's an insecure connection so Chrome mobile may (in the interest of protecting your phone) refuse displaying 'mixed content' by default. It will also give a 'page insecure' warning in the address bar.

While in the other thread the images come from https://i.imgur.com and so it's secure.
bogo uses Linux mainly, otherwise Firefox. He seems to be hitting the problem with large externally-hosted images.
It might be associated in some way with the 'traffic optimisation' changes rather than the software upgrade -
although I've not hit any problems with images, I've noticed some oddities - e.g. when I use (Google/ Chrome) autotranslate from a foreing language on a page that's secure, it stops being secure in English :confused: (presumably the encryption stops working?)
 
bogo uses Linux mainly, otherwise Firefox. He seems to be hitting the problem with large externally-hosted images.
Well I can't debug him remotely :D what I can see is -- on Margot's thread I get the same result as in the screenshot he posted, 100% of the time. The photostory thread works.

And the difference is that Margot's thread uses non-HTTPS embeds Imixed content, secure and insecure)
and the photostory uses HTTPS (only encrypted content)

So a simple test for him would be to check the photostory and see if he can get the images there to display.
compare https://www.cruxforums.com/xf/threads/news-from-margot.4343/page-35 to https://www.cruxforums.com/xf/threads/the-soppy-sentimental-photostory-thread.8280/
 
That has always happened every now and then and is probably just when the number of available connections is exceeded or so.
Yes it is. And because you know about these stuffs well, I can add that we might be able to deal with the problem by applying some aggressive caching to the dynamic contents, like the front page, for example. If it's not the case where MySql is inappropriately tuned, even such an extremely short-lived cache which expires in a few seconds may do wonders here.

bogo uses Linux mainly, otherwise Firefox. He seems to be hitting the problem with large externally-hosted images.
It might be associated in some way with the 'traffic optimisation' changes rather than the software upgrade -
although I've not hit any problems with images, I've noticed some oddities - e.g. when I use (Google/ Chrome) autotranslate from a foreing language on a page that's secure, it stops being secure in English :confused: (presumably the encryption stops working?)
I'm also on Linux 99% of the time I visit this site :) (By the way, Firefox is a browser like Chrome, while Linux is an OS like Windows, to be exact.) I don't have the slightest idea what's happening with auto-translation, however. So if you see it happenning again, please let me know how I can reproduce it on my end so I could test it.

I think @malins is right in assuming it is a mixed content problem caused by images referenced with the insecure protocol (i.e. http).

And I suspect that it could possibly have been aggravated by a recent upgrade of Chrome which started blocking insecure images (as well as scripts) from the version 81 and upward.

For those who are interested in reading the details, please read these articles:
The problem is that those particular images cannot be "upgraded" because the host (redfeline.com) does not support HTTPS. So unless either the host can move to HTTPS like we did, or the user can some how allow less secure images by using a less strict browser or changing a relevant option, there's really no remedy for that.

It's a problem which was bound to happen sooner or later as the internet has been trying to deprecate the non-secure protocol for quite sometime. All major browsers have been introducing increasingly stricter policies regarding insecure/mixed contents, and they will eventually just block them to force people to migrate to HTTPS.

So, I believe the ideal way to handle this problem is to contact the owner of Red Feline website and encourage him or her to switch to HTTPS like we did.

According to Google, over 90% of websites people open with Chrome are already on HTTPS. So, it's just a matter of time before their images won't be able to be displayed on most websites if they don't upgrade anyway.
 
Last edited:
we might be able to deal with the problem by applying some aggressive caching to the dynamic contents, like the front page, for example.
What do you think about the debug output at the bottom right of pages (CSS class p-footer-debug)?
It gives us info about the current controller class and method, and some statistics, such as

Queries 8 Time 1.8094s Memory 17.94MB

Usually collecting debug info comes with some performance overhead and should be turned off on production systems,
maybe this is an element that can be switched off to gain some performance.
This kind of information is really only necessary if the forum is currently being debugged,
such as to see whether some plugin generates an awful amount of MySQL queries or so, or something is eating memory.
 
What do you think about the debug output at the bottom right of pages (CSS class p-footer-debug)?
It gives us info about the current controller class and method, and some statistics, such as

Queries 8 Time 1.8094s Memory 17.94MB

Usually collecting debug info comes with some performance overhead and should be turned off on production systems,
maybe this is an element that can be switched off to gain some performance.
This kind of information is really only necessary if the forum is currently being debugged,
such as to see whether some plugin generates an awful amount of MySQL queries or so, or something is eating memory.
I agree with you that such metrics usually incur some overhead and better be disabled in a production website. But I don't know if the overhead in this case is severe enough to cause such problems.

Personally, I doubt it could be the case, especially with a traffic like just about 40 queries / second (it reads 8 queries in 0.1983s on my end). But who knows, I don't have that much confidence in MySQL and it could be the case it's running on a potato server.

Nevertheless, I agree that there is no reason why such debug information is enabled on a production server and it's better to eliminate whatever overhead it may incur.
 
question ... is that total queries on the server ... or the queries used to just generate that one page, for that one client ... there might be many page requests within one second.
(EDIT) I wrote something about crucifying the XenForo developer but I just changed my mind. Yes, I think you have a point. I somehow assumed the displayed time could be an average metric but now I realize that it can very well be simply the time taken for the page to be rendered.

Actually, I now think your explanation is more plausible because it constantly shows 8 queries which can hardly be the case if it was a metric for the total queries count.
 
Last edited:
Back
Top Bottom