SJ77

Btw, I haven't checked what happens in the gallery. You might want to check that area, too.

Here are the steps to reproduce:
Create a new file in the Downloads app (in a category that allows screenshots). Upload any random file in the first step (it doesn't matter). Upload the image from the ZIP file attached to this post as a screenshot. Enter a random title (it doesn't matter). Re-upload the image from step #3, but this time as an attachment in the Description editor. Submit the file.  
Once the file is submitted, compare the screenshot to the attachment in the description, and you'll see the images are rotated differently (and rotated wrongly in both cases, too).
test_image_rotation_bug.zip
 
P.S.: The test above was done with both GD and ImageMagick. Same result in both cases.

Just my 2cents: if you have any AV (or mod_security) configured, you may want to exclude these large files (if they are not exluded by default), otherwise they can take up a LOT of memory while processing them. 

It depends. You ideally don't want it so small you end up having to make a crazy number of requests for larger uploads. For a 1GB file, you'd need to make 100 requests to finish uploading a file, which is a decent bit, but as this isn't sustained traffic it's not the worst thing. People with faster connections may hit you with a large spike of requests when uploading larger files, as a gig line can push 10MB of data in no time at all, so just need to make sure you can handle that if you want to keep it at a lower value.
You could increase it a bit if you wanted to reduce that overhead a bit, as 10MB is pretty low. But unless you're noticing performance issues, particularly when uploading files, it's probably not a big concern.

disregard, I now believe it was a "proxy" setting in firefox that was causing the issue.
Thank you

After all these years users with iphones uploading pics are still upside down, just happened on my forum again and exif is enabled with PHP 8.0
 

Thank you for bringing this issue to our attention! I can confirm this should be further reviewed and I have logged an internal bug report for our development team to investigate and address as necessary, in a future maintenance release.
 

I would specifically request that optimizations be made to the getItemsWithPermission() function
Seems like all things and kitchen sink at the moment. Perhaps there is opportunity for SQL optimizations here?
Queries relating to this function are currently the slowest on my site. they regularly show in my slow query log.

One item holding me up from moving to the cloud hosted plan is the lack of SFTP access. Even if this was to a single folder in the uploads directory or something, it's all I need. My feature request is some type of uploading capability. Perhaps it's there and I don't know it.
I need this because I upload advertising banners to a nonstandard directory, then use custom code to call them. This way, no ad blockers block my ads. 
I also use SFTP to upload the audio MP3 of my articles, then put them into the articles with the code below. I would attach the MP3, but this displays how many times each article is listened to for some reason, and the accuracy of it is questionable, in addition to the fact that it's a stat that nobody needs to see.
<audio controls="" data-audio-embed=""> <source src="https://example.com/my-article-audio.mp3" type="audio/mpeg"> </audio>  

To be honest, I kind of wish this thread wasn't here with debate starting etc. I basically just wanted a support ticket with IPS. I thought this is how it worked now days.

My idea for a ChatGPT integration:
autoregister new users let them create content that fits my forum topic auto moderate my forum If you get this done I will asap switch to a cloud plan.

 

I don't know about you, but I'm wondering if it's perhaps overdue that IPS got stuck into a huge performance and speed based review and implement it in the next medium sized update! I think a lot has changed in recent years in relation to site performance. Core Vitals in Lighthouse as many admins will be aware are now much more important in terms of SEO ranking, performance isn't just for the obsessed enthusiasts any longer!
Google Search Console tells me that on one of my IPS Communities, my Mobile Pagespeed test result sucks, just 28%-30% at present, so I started checking a few things under the hood. I like to keep my sites high performing, so I was quite surprised it was so low. 
One thing I found out in minutes looking at the optimisation list and then checking my IPS configuration is that if you have the IPS Lazy Loading feature enabled, it doesn't apply to your uploaded images in Advertisements or Post Author Member Group Icons. Both are shown in my topic view so I get scored down on mobile in particular.
I think it's now time that Native Lazy Loading was offered, either as a choice or as the default if enabled.
Also where possible IPS should support Local WebP conversion via the Task Manager.
I know some of us sometimes bash IPS for not delivering more obvious updates like Gallery, Blog, Pages and Commerce improvements for the sake of instead quick things like disabling features but IPS4 is quite a huge behemoth these days, no matter your pricey VPS, CiC, use of Redis pools etc.
More performance improvements I'd like to see considered include...
(Now admittedly I'm basing this list on a similar big audit that another website framework I invested in conducted and implemented last year. These are the options you can now simply toggle off/on as a result in their Dashboard. They can make a huge difference to bloat.)
Dynamic Modular Framework
Enable this to allow the IPS Framework to only load code that are actually used on the page, and process the logic for the features in actual use.
Dynamic CSS
Dynamic CSS could greatly reduce CSS file size by dynamically generating only the styles and assets necessary for the features and modules you actually use. This eliminates most if not all CSS file bloat and greatly improves load times.
Dynamic Icons
Font Awesome is 2 generations behind. The FA4 icon font could be broken up into various subsets. These subsets are loaded only when needed based on the modules and features used on each page. If you need access to the entire icon font on all pages (for example, if you are using lots of FA icons in a custom theme or want to allow members to insert them via the post editor), then you could disable this option and load the entire icon font library on all pages.
Load Dynamic Stylesheet In-line
This option dequeues the IPS style.css file and prints the needed contents in-line. This removes a render blocking request and improves the PageSpeed scores of individual pages. However, it also prevents the style.css file from being cached. Since the stylesheet is often smaller when dynamic or critical, it could be recommended to keep this option enabled. Accepted there are some potential CSP-related security concerns with inline CSS an or JS.
Critical CSS
Critical CSS greatly improves website load times and Google PageSpeed scores by deferring non-critical styles and eliminating render-blocking CSS requests. Critical CSS often covers what most users see when first loading the page, so called 'above the fold' content.
Critical Threshold Height
When Critical CSS is enabled, IPS would determine an "above the fold threshold" and defer all styles for elements below the fold. However, this threshold is just a estimate and can vary on different devices. Increasing threshold height will deffer fewer styles, resulting in slightly slower load times but less of a chance for Cumulative Layout Shifts to occur. If you are experiencing CLS issues you can then increase or lower the threshold 'height' via a drop-down low, medium, high setting.
Dynamic JavaScript Libraries
When enabled, IPS would only load external JavaScript libraries when they are needed by a specific sections or modules on the page. This removes unused JavaScript from the main scripts bundle and improves load times.
Improve Google Fonts Loading
Enable caching of Google Fonts and load them inline. This reduces render-blocking requests and improves page load times. Also presents an opportunity to switch to the Google Fonts API, to offer a dynamic dropdown list to allow admin to choose from all available GFonts. The Task Manager could check for and cache new versions of selected fonts in use weekly.
Limit Google Fonts Support For Legacy Browsers
Enabling this option will lower the size of Google Fonts and improve load times, however it will limit Google Fonts support in some very old browsers. You can turn this off to increase support for losers insisting on their older insecure browsers at a slight cost to performance.
Defer jQuery And jQuery Migrate
When possible, jQuery and jQuery Migrate could be moved to the body to speed up load times. If a third party plugin registers jQuery as a dependency, it would be moved back to the head.
Enqueue jQuery Compatibility Script
Some third party scripts may be incorrectly enqueued without declaring jQuery as dependency. If jQuery is deferred, these scripts could break. If you are experiencing console errors after enabling the "Defer jQuery And jQuery Migrate" option, you can enable this option, which will load an additional compatibility script that will attempt to solve the issue. (You can turn this feature off if everything is working fine without it).
Defer Additional Third Party Scripts
Warning: This can cause JavaScript errors in some cases and should be used with care. When enabled, scripts registered by plugins and themes will be deferred to improve performance and jQuery will always be loaded in the body.

To be honest, I kind of wish this thread wasn't here with debate starting etc. I basically just wanted a support ticket with IPS. I thought this is how it worked now days.

4.4.9 is just slightly over 3 years old.
  
Previous was 7.2 now 8.1
  
Nope was running a version of Maria DB that is only a year old. Still using the same Maria DB
  
Spent $$$$ (a lot) working with top IPB developers to optimize every single bit of code for both PHP 8 and IPS 4.7.7 (one by one), It very expensive and took months, I am confident I am working with the best people in this regard. 
  
I am 100% confident that I applied the settings correctly.  I did preciously what everyone else is doing. The problem is that users who don't know how to clear their browser cache end up being locked out of the site. I believe others don't notice because only .001% complain. Since I have millions of visitors, I get hit with lots of feedback.
  
I have been a system administrator and developer since the early 2000's. But I also work with other very skilled system administrators and developers. You should know I am not a hobbyist. We are operating a registered corporation and have been in business many years.
  
I agree, there is no point passing blame around. My thread here is not intended for that. The goal is for solutions only. I want you to know that you have been very generous and kind with your assistance to both me and many others around this community. I really appreciate that Randy. Thank you for being so amazing.

While I am upset that the upgrade has caused (and is still causing me) a lot more stress than ideally I would have wanted, I have nothing but gratitude for the software, the entire IPB team and this community. I feel very blessed to have a business that runs on IPS software

All support on the forums do get our staff's attention. If anything cannot be handled here then we sometimes convert them to a ticket via email.

Support is now started through the community here. The "Help & Support" forum here is managed by our staff and this is the only official support forum which we have. We no longer have priority support but everything else functions the same, our staff still has access to the Client Area to securely access your access details to your community, we can escalate items to tickets with our developers (if needed), etc... 
The reason for this change is that we receive a super high percentage of requests/questions which are extremely similar in nature and our clients could benefit from searching a common area for those answers. Additionally, other clients can chime in and report their experience with a particular issue or solution, etc... 
Many years now we have been commenting to our clients how great communities are for support and never have been using it to it's full potential so we are now 🙂 .
Just like in a ticket though, we will move items which fall outside our scope of support to the appropriate forum, like performance on your server, to a more suitable forum. 
Keep in mind, all support on the community is public now, that is the benefit of doing so.

After days of experiments and hours of head scratching, I can only conclude that 4.7.7 is more resource intensive than 4.4.9

I suppose I wrongly assumed that it would be more efficient and not less. However, in retrospect, I guess since it is more complex, and doing many more things, it makes sense.
Before upgrading I believe it's important to make sure one has sufficient additional resources available. I was not prepared to go from average load of 8 to average of 25.. shocking!!!  Now my site is really struggling.  There really isn't anything broken per se, thus my next option is to perform emergency migration to a more powerful server.

Again, thank you all for your input. I greatly appreciate it. 🙂

CloudFlare has been proposed in the deprecation topic as an example of how easy it is to setup a CDN to do page caching. It is not the only solution by any means. If you cannot work in CloudFlare, there are countless solutions out there. You will want to find one that suits your needs 😉 .

@SJ77
Just a hypothesis - you are updating from 4.4 to 4.7.7
I'm not sure from which version service worker running was introduced, but I think it was after 4.5.
Now in many tests - your landing page lags over 12 seconds because of this service. Your other pages load 10x faster..
It is possible that this new service does not cover your landing page, so this rather long wait for your landing page causes a high level of load on your server.
Try it - shut down your landing page for half an hour and monitor server load...
You might need a landing page edit because most of your users are guests, and that's a lot of high-latency requests...

It's not really a fair comparison. The reason people use our site goes away without 3rd party resources. (it's basically like having a car without wheels)

With that said, I tried it and the load did go down, but still higher than on 4.4

All of my 3rd party resources were inspected (and or improved) by a really incredible developer prior to upgrade.
At this time, as aforementioned I don't see any errors in the logs.

The main consumer of cpu is mysql so I started checking slow query logs and monitoring process list
The main things are all long running default IPS stuff that aren't anything to do with 3rd party plugins. (such as the example I gave above that was running for over 10 seconds)
I also followed directions on setting up cloudflare for guest caching. I think it might have helped a little but wasn't a golden ticket.

Thank you for your response 🙂

You would have had to move from php7 to php8 to be upgraded. I would highly look at optimizing your PHP as it’s likely not designed for high performance with an out of the box setup. If you’re running php-fpm, it has its own config that needs optimizing. 

That was a lesson I learned the hard way more than a year back. 

That was my issue. Disabled and my phantom msql load gone. Not sure if it’s a conflict or what and have not had time to look deeper. I also enabled slow query logs and nothing really came up. It was also preventing some of my newsletters from going out. I thought it was that app at first. 
Just saying, shut it off and check as well as other apps.

I would actually recommend disabling all 3rd party as a first instance, and test without. Until you do this, you are just going to be guessing at what may be the cause. By doing this, at least you can either rule it out, or rule it in. If you then find it is, you can start enabling things one at a time till the issue returns.
One thing I would advise on checking first of all, is that none of the table types changed. Ensure they are all INNODB on your server.

That means your server’s PHP8 is not working correctly. You’ll need to work with your hosting provider. PHP is most likely not compiled correctly. 

If that script cannot run, it is absolutely a server problem. 

Create one with the only code in it being what is shown:
<?php define('RECOVERY_MODE',True); It should exactly be named and uploaded as:
constants.php
Once it’s uploaded, open a new browser and login to ACP.  You’ll see a message about things being disabled.
Once done, you’ll hopefully be in ACP with a message saying turn recovery mode off.  Simply delete that file to do that.  Then you’ll just need to figure out which third party resource is breaking your site by turning them on one by one till the site breaks again.
Rinse and repeat.  🙂