Clover13

2 minutes ago, Randy Calvert said:

Cloudflare won't have anything to do with this.  The 403 is coming from origin (S3 in this case).  So you want to focus your attention there.  

Have you looked at your bucket policy settings for each site?

https://stackoverflow.com/questions/26691286/amazon-s3-bucket-returning-403-forbidden

Yes, I compared both but will look again.

What's odd is this isn't global for all files, it's just certain files.

Just to add CF's Hotlink Protection is OFF

Not sure what's going on exactly, but it seems on two different IPS sites I run some images retrieved from S3 via CF ala a CDN CNAME (i.e. cdn.site.com) return a 403, however the majority return just fine.

File types that return 403:  ICO, SVN

One site returns SVN fine, both sites 403 for ICO

Both sites return JPG, PNG, GIF fine.

I've compared htaccess, S3, CF configs from both sites and haven't spotted anything different between them.

Anyone have any ideas?

39 minutes ago, Marc Stridgen said:

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.

 

Thanks @Marc Stridgen, is there any workaround to get a custom FURL in place?  Can I add it to the DB via a query?

FURL:  recent-topics

Real URL suffix:  app=core&module=discover&controller=streams&id=6

TypeError: Unsupported operand types: string + int (0)
#0 /home/*****/public_html/system/Dispatcher/Controller.php(107): IPS\core\modules\admin\settings\_advanced->furlForm()
#1 /home/*****/public_html/applications/core/modules/admin/settings/advanced.php(38): IPS\Dispatcher\_Controller->execute()
#2 /home/*****/public_html/system/Dispatcher/Dispatcher.php(153): IPS\core\modules\admin\settings\_advanced->execute()
#3 /home/*****/public_html/admin/index.php(13): IPS\_Dispatcher->run()
#4 {main}

FWIW, got it sorted it out forcing a manual update of the local parameter initially.  Something with CloudLinux not behaving/updating properly, host had to toggle from PHP Selector mode to MultiPHP Manager mode and things synced up properly when he went back to PHP Selector mode.

OK thanks guys I'll follow up with the host

The other odd part is the ips4.php shows full compatibility, would this not be flagged/caught as part of that compatibility check or as part of an upgrade compatibility check?  Or is there no way to do that?

1 minute ago, Jim M said:

Unfortunately, your phpinfo screenshot there is clearly showing it as well which is your local configuration so that would indicate something on your server is altering it.

Right but the question is why would the ips upgrade reference the local when there is a master override for the selected version (8.1)?  Is that normal for local to be referenced or is there different scoping whereas local needs to be 8.1 and operates differently than master as 8.1?  I had though the master was the primary and superseded the local when it is specified.

1 minute ago, Jim M said:

You would want to contact your hosting provider for assistance. Unfortunately, this is not something our software is doing but rather your server's PHP configuration.

I have, they are reviewing it but they also don't understand how the IPS upgrade process is referencing something other than what is specified when the 8.1 php version is selected

I have PHP 8.1 selected and the IPS ACP shows 8.1 running and ips4.php shows 8.1, however the upgrade failed due to 

We were unable to start a PHP session. You will need to contact your host to adjust your PHP configuration before you can continue. The error reported was: session_start(): Failed to read session data: files (path: /var/cpanel/php/sessions/ea-php72)

Not sure why I'm getting an error referencing the local value instead of the master value.

Why is the upgrade referencing 7.2?

3 hours ago, cofee_iv said:

Now that Paypal has shown their colors, I am looking for this, as well. I plan to remove my PayPal app from the forum.

Not sure what you mean about Paypal showing their colors, but Paypal owns Venmo as well.

2 minutes ago, teraßyte said:

777 is what IPS suggests, but depending on your server's setup the required permissions might be different. As long as your tasks are running properly you're good.

Got ya, yeah they're all running ok as far as I can tell and per the task logs and IPS dashboard status.

18 minutes ago, teraßyte said:

Yes the task.php file needs to be 777 in order to run properly. And the key at the end (set of numbers and letters) is a parameter that gets passed to the backend code when it runs. Leave everything as is and you're good. 👍

Mine at /applications/core/interface/task/task.php is 744

EDIT:  Nevermind I see the path, had to scroll

FWIW, my site converted smoothly, but not running many third party apps/plugins.

Invision Community
Version 4.7.2.1

PHP
Version 8.1.10

Database
Version 10.6.10-MariaDB

I did get a notice on the GMP PHP extension missing that I'm looking to get resolved with my host. 

Quote

Web Push Unavailable
Unfortunately, your server does not have the GMP PHP extension installed. Please contact your host or consider switching to Invision Cloud Community which has Web Push Notifications enabled.

It does bring up the same question as @Dexter_X regarding what the list of PHP modules required (or optionally beneficial, i.e. caching) are.

P.S.  I'm only asking because I saw this behavior on mobile only that suddenly arose yesterday and it was directly related to a CF configuration issue of a beta feature.  May not be your issue at all, but figured I'd throw it out there as I'm not clear on how your CDN links were working without CF unless you weren't routing your CDN address through CF.

Is your CDN endpoint server (AWS) not routed and cached through CF (i.e. I have a CNAME for my cdn.* to point to the AWS S3 location)?  When you disable CF, what happens to your CDN routing?  Or are you controlling that on your server side?

I hosted my media on S3 and have Cloudflare as well, but am noticing the last month+ that Bingbot has become very aggressive in crawling historical data (I suspect given it isn't hitting the CF cache but rather pulling from S3 directly).  Anyone else experience the same or have any idea what may be going on?  I've been in S3 for a few months now and this spike just started last month from what I can tell from the AWS logging I'm doing and querying via Athena.  

Bingbot is running 43,186 S3 GetObject events per day ranging from 4-9GB in total per day from this bot.

Amazonbot is also starting to increase recently (from what I understand for Alexa, so perhaps blocking Amazonbot won't hurt SEO much).   I'm seeing 6,187 S3 GetObject events per day ranging from 2-6GB in total per day from this bot.

Meanwhile Google is running 7,915 S3 GetObject events per day all less than 1GB for months with the exception of one day.

Ultimately it's not some insanely high costs, but certainly Bingbot at 40K GetObject requests a day vs Google at approx 8K is significant.  Not sure if this will slow down as Bing gets better indexing.  Also not sure if porting to S3 caused this gross reindexing to go on or why it's happening at this level.  I tried to adjust the crawl delay/rate and de-prioritize it both in robots.txt and Bing Webmaster tools but it had zero effect.

My site uses an Activity Stream as the primary home page.  We have a site defined default stream that members can override (which updates the link to the far right of the bread crumbing on the page as an effective "quick link" to it). 

Is there any way a member can define their default Activity Stream and have the site's default homepage land there for them as opposed to bookmarking or doing a multi-step process of visiting the default homepage and then clicking their default stream on the top right?

On 6/8/2022 at 6:45 AM, Andy Millne said:

Some communities restrict their default member group from being able to send private messages. You can do this and then the system can move them to a higher group with elevated permissions once they have posted x publicly approved posts.

We did this and it has been effective in stopping them.  It's very rare for us to see them post enough content to gain PM access before the content gets flagged for clear boosting (we know the game they play well).  However our members are active and help mitigate this issue by reporting bad actors that slip through the cracks.

Or you can hide them, that's generally what we do on my site, so mods have awareness/context and it remains within in the topic for mod view but not user/member view.  Also allows you to restore them (unhide) as needed.

OK thanks Jim, been struggling with trying to solve this for a clear line in my community for some time, and no matter which side I choose as a default the other half isn't happy about it (but also too lazy to set their own default, maybe partially out of lack of technical understanding or out of principle).  Good times hah

I know Members can set their own Activity Stream default configuration, but is there a way I can have this set automatically for them based on selection (i.e. they get a prompt to select their choice) or on a Member Group level (where I can somehow give them an option to choose and place them in a corresponding Member Group).

Ultimately this use case comes down to two distinct groups of Members wanting different default views but not wanting to do the work to set it, so I want to force them to select an option.

Edit:  Updated title

7 hours ago, Andy928 said:

Ok guys I found the source of the problem.

This is el defo an issue with Invision backend. This is also why this invisioncommunity forum does not experience this issue. They have the "Paste as rich text initially, but allow user to paste as plain text instead." enabled. Anyone else experiencing the duplicating image issue like us is because they have the "Always paste as plain text" enabled. I was able to replicate this in all the browsers I tested.

Invision can you please fix this?

Thank you.

 

Nice find and troubleshooting! 👍

Awesome!  Thanks for the feedback all, and great example layout @Daniel F!  I was playing around with the REST API last night and able to create topics through it via an apikey, but I'll look into the Zapier approach you detailed and give that a go!