Issue Redirection from WooCommerce fail - since update 18.0.61 #2?

[andreios]

Server operating system version

Ubuntu 22.04

Plesk version and microupdate number

18.0.61 #2

After update #2 there is a problem in two of my stores, the stores have multi currency enabled and WooCommerce adds an argument to the URLs e.g. /?v=3a52f3c22ed6
It only happens on the start page. I can't rule out other sources here, but it only seems to occur since update #2 today. It doesn't always happen, sometimes it happens, sometimes not, it doesn't seem to have a reason.
nginx cache is off, proxy is off.
Every time it fails, the following error appears in the log:
child 15256 said into stderr: "PHP message: PHP Warning: Undefined array key "host"
The header contains the following:

location https:///

It should look like this:

location https://domain.com/?v=3a52f3c22ed6

The Browser says:

ERR_INVALID_REDIRECT

These are the only shops on my Server with multi currency enabled. Other domains seem to have no issues.

 

[Kaspar@Plesk]

Does upgrading to Plesk 18.0.61#3 fixes to issue?

 

[andreios]

After update #3 it doesn't seem to happen anymore, at least so far. I will keep an eye on this.

 

[andreios]

Te problem still exists, on multiple domains, causing different issues. I could rule out any plugin or theme.

 

[Kaspar@Plesk]

Has HTTP/3 been enabled on the server?

 

[andreios]

Yes, HTTP3 is enabled for both panel and nginx. But I have tried yesterday without it and the issue still happened. But from the time frame it would fit, I enabled HTTP3 yesterday. I have disabled it again. No issues so far and nothing in the logs. But needs more time to be sure.

 

[andreios]

The problem seems to be related to HTTP3. There are no more problems to be found in the logs. People who have reported problems say that it is working again.

 

[Kaspar@Plesk]

I *think* this might also be related to the same issue mentioned in this forum thread, caused by the fact that the HTTP_HOST server variable isn't when using HTTP/3. But I am not 100% sure, as it's difficult to say for sure based on the information provided.

If you're up for it, you can test this by enabling HTTP/3 again and adding fastcgi_param HTTP_HOST $host; to the /etc/nginx/fastcgi.conf file.

Note however that this is likely not a sustainable solution as the fastcgi.conf might get overwritten when Plesk gets updated. But that way you'll know for sure that it's an HTTP/3 related issue.

 

[andreios]

Yes, I thought it must be the HTTP_HOST Variable, but I didn't know nginx has an HTTP/3 issue with it.
I added

fastcgi_param HTTP_HOST $host;

and tested it again.
No issues and nothing about 'host' in the logs.

 

[Kaspar@Plesk]

So if enable HTTP/3 again and add fastcgi_param HTTP_HOST $host; you have no more issues? Is that correct?

 

[andreios]

Actually my browsers did still use HTTP/2. I tested on HTTP/3 Check everything was HTTP/3.
I tested also with a .php file with $_SERVER, if HTTP_HOST was set, it was. But my browsers were using HTTP/2.
I restarted nginx, but still both of my browsers were using HTTP/2.
Just five minutes later, I did nothing, one of my browsers started using HTTP/3.
In this browser, my .php test file says HTTP_HOST is not set. Also, I see the issues appearing again in this browser.
But fastcgi_param HTTP_HOST $host; is definitely set in /etc/nginx/fastcgi.conf the whole time.

 

[andreios]

Right after this, I have disabled HTTP/3 and enabled it again.
I have installed Opera, and Chromium and Opera use HTTP/3 right now, Firefox still tries with HTTP/2, don't know why. But, HTTP_HOST is there. Have I just overlooked it and it was there the whole time?
But the issues are also diapered. No issues anymore the whole time.
Also, the browsers sometimes switch randomly back to HTTP/2.

 

[Paulo_o]

I have installed Opera, and Chromium and Opera use HTTP/3 right now, Firefox still tries with HTTP/2, don't know why. But, HTTP_HOST is there. Have I just overlooked it and it was there the whole time?

Do you use nginx with worker_processes greater than1?

 

[andreios]

It is set to:

worker_processes  1;