Applicable to:
- Plesk Obsidian for Linux
Symptoms
-
Websites show the following error:
CONFIG_TEXT: AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/example.com/php-fpm.sock|fcgi://127.0.0.1:9000/var/www/vhosts/example.com/httpdocs/public/index.php)
CONFIG_TEXT: Internal Server Error
-
Apache has been updated to a new version recently:
Ubuntu 20.04:
# grep 'status installed' /var/log/dpkg.log | grep apache2:amd64
2021-09-27 12:46:57 status installed apache2:amd64 2.4.41-4ubuntu3.5Ubuntu 18.04:
# grep 'status installed' /var/log/dpkg.log | grep apache2:amd64
2021-09-28 06:25:55 status installed apache2:amd64 2.4.29-1ubuntu4.17
Cause
This is an Apache bug which arose with Ubuntu packages update and later fixed in the following versions:
- Apache for Ubuntu 20 version 2.4.41-4ubuntu3.6.
- Apache for Ubuntu 18 version 2.4.29-1ubuntu4.18.
Plesk uses Apache packages from OS repositories and does not control their version updates.
Resolution
Update Apache packages to fix the issue:
-
Go to Tools & Settings > Server Management > System Updates and click Recheck now.
-
Once updates are detected, click Update all.
The issue should be fixed, verify that websites are back online.
-
Connect to the server using SSH.
-
Execute below command:
# sudo apt-mark unhold apache2
-
Apply steps 2-3 from the resolution.
Comments
318 comments
Thank you very much for the quick workaround. I was still trying to use the 1st workaround but then it updated by the time I figured out how to SSH. AWS Lightsail actually makes it quite easy. Well we live and we learn hey...
I just replaced PhysicalHosting.php, rebuilt all configs, removed the hold from apache2 and ran the updates to the version that caused all problems this morning. Looks fixed! Everything works like it should. Thanks for this permanent fix!
//I applied the workaround, but only some of my sites came back up. Some still show the 500 error. I tried switching to FastCGI on one of those with no effect. The error shown for those sites in the log is still AH10292....
Solved: For those domains I had special settings set in Plesk to include in httpd.conf which were overriding the fix. Someone mentioned changing the proxy URL (/// to /) for the domains and the "Apache and Nginx Settings" were where that needed to be changed, too...
@Nick H - The new fix needs to be run as root. If you want to use sudo, make sure you add it to the tar command:
ive also just done the fix and confirm the actions above Johan Adriaans worked (so far) like a charm. many thanks
Urgh, does anyone know how I can implement a fix using Plesk for Web??? I'm all at sea here....
Well the user that reported it did not use the official Ubuntu packages but rather the repository of Ondrej Sury which maintains packages of PHP and Apache for Debian/Ubuntu and releases updates more frequently since it appears he also works on apache itself. He reported it to Ondrej Sury on 22nd so sure enough Ondrej fixed it directly at apache but on time for Ubuntu's new release.
Benjamin Weßel,
I cant read the new code so, have you reviewed the use of the variable $VAR->domain->physicalHosting->fpmSocket on all the sourcecode?
For example it is been used also on /opt/psa/admin/conf/templates/default/domain/service/fpm.php (used by nginx domains not using "Proxy mode")
It could generate
instead of
Thanks
Had this scratched my head for ages and found this page. However, I'm getting the below when running the command:
Saving to: ‘PPPM-13232.tgz’
PPPM-13232.tgz 100%[=================================================>] 33.40K --.-KB/s in 0.002s
2021-09-28 12:41:24 (19.4 MB/s) - ‘PPPM-13232.tgz’ saved [34203/34203]
PhysicalHosting.php
tar: PhysicalHosting.php: Cannot open: File exists
tar: Exiting with failure status due to previous errors
Running wget command now (as of 13:47 28-09-2021) 404 not found
wget https://support.plesk.com/hc/en-us/article_attachments/4407402583954/PPPM-13232.tgz && tar -xvf PPPM-13232.tgz -C /usr/local/psa/admin/plib/Template/Variable/Domain/
--2021-09-28 12:46:44-- https://support.plesk.com/hc/en-us/article_attachments/4407402583954/PPPM-13232.tgz
Resolving support.plesk.com (support.plesk.com)... 104.16.53.111, 104.16.51.111
Connecting to support.plesk.com (support.plesk.com)|104.16.53.111|:443... connected.
HTTP request sent, awaiting response... 404 Not Found
2021-09-28 12:46:44 ERROR 404: Not Found.
Why there`s a different workaround than 1 hour? I already downgraded my Apache Version and set it to hold. What to do best?
Thank you! You saved me a lot of work!
Workaround changing it to "Run PHP as FastCGI Application" worked also for me. But your solution is much better. Thx!
Thank you so much for the quick support
Can you explain how to do that? I'm not good at understanding these things but my website has been down for several hours.
On AWS Lightsail the wget command throws 404
https://plesk.zendesk.com/hc/article_attachments/4407403144722/PPPM-13232.tgz - 404
This workaround works easily.
If asked to keep log rotation confs, answers NO
@Thorsten Bachner
Downgrading and locking apache is not future proof. It is a temporary fix. The new version is a permanent fix.
So I've just installed the new Apache2 Version from Ubuntu which has not been released officially yet and without the new workaround (to PhysicalHosting.php) this works as it should. New Version for Ubuntu 20.04 will be 2.4.41-4ubuntu3.6
So once this releases officially and plesk updates, it will be fixed permanently.
(I've upgraded from the downgraded apache, so it's 2.4.41-4ubuntu3.5 -> 2.4.41-4ubuntu3 -> 2.4.41-4ubuntu3.6)
This is getting difficult to keep up with.
For those of us that ran the downgrade and hold commands earlier, can someone clearly explain what we need to do now?
I don't see an unhold update above nor do I understand whether we unhold apache and then manually re-run the update or wait for plesk to do it.
It would really help to get a clear set of instructions for those who actioned the downgrade and those who didn't.
I changed everything to select "FastCGI application served by Apache" and it all works again. Lost close to a day with a number of websites
Just to clarify that the below worked for me:
NOTE: the sudo commands at the beginning of the line and also before the tar command.
The second command above to reload configuration also needs the sudo command:
Alright guys, here's the workaround if you are using Ubuntu 20.04. Im not sure about Ubuntu 18.04, but this should work.
1) Login to your plesk dashboard
2) Click on your domain name you would like to do these changes.
3) Go to PHP Settings
4) On the top of the page where it says "run PHP as" select "FastCGI application served by Apache"
5) Scroll all the way to the bottom and click Apply and OK
If the above method doesn't workout for you, try the next solution which is,
1) Again, Go to PHP Settings
2) On the top of the page where it says "run PHP as" select "FPM application served by nginx"
3) Scroll all the way to the bottom and click Apply and OK
4) Now, Click on Hosting and DNS
5) Click on Apache and nginx Settings
6) Scroll down to nginx settings
7) Uncheck "Proxy mode"
8) Hit Apply and OK
Enjoy :)
All we can do is to wait patiently until the Plesk team releases a stable fix for this issue, if you dont know what you are doing with SSH, stay away from it as it can mess things up, just a heads up, unless you really have experience with SSH and clearly know what you are doing :)
Had just started receiving calls from clients. Thank-you for the quick fix
@Thorsten Bachner
1. Remove the hold status for apache2
2. Update & Upgrade
3. Replace the updated version of PhysicalHosting.php
4. Rebuild webserver config
Thx a lot! I'm working with a DO droplet that included plesk, and the quickfix worked great. Is there a newsletter or something where I can be updated in realtime of these events?
Felix Gade
Thank you for that. It all worked apart from step 3 where I got a 404
To fix that, I ran the command again and it worked.
Rechecked which Apache version is installed and it's back to
2021-09-28 14:25:19 status installed apache2:amd64 2.4.41-4ubuntu3.5
@Hostworks Admin
Is it possible that you have some websites set up with FPM and some with FastCGI?
Today we woke up to more than 35 sites down, I was lucky enough to find this post, the previous version didn't stated to run the commands as sudo but we made it work anyways. Thank you and please keep us posted.
Felix Gade
Tried the latest fix and getting this missing trait error when running
Please help.
Please escalate the help... I got flooding server errors from the app api impacting lots of users
[Tue Sep 28 06:34:45.715574 2021] [proxy:error] [pid 14836:tid 140299611342592] [client 47.145.99.21x:49404] AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/...
[Tue Sep 28 06:34:45.850470 2021] [proxy:error] [pid 14836:tid 140299602949888] [client 47.145.99.21x:49406] AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/..
[Tue Sep 28 06:34:45.974177 2021] [proxy:error] [pid 14864:tid 140299720447744] [client 47.145.99.21x:49408] AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/...
[Tue Sep 28 06:34:46.158366 2021] [proxy:error] [pid 14864:tid 140299661698816] [client 47.145.99.21x:49410] AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/...
[Tue Sep 28 06:34:46.529260 2021] [proxy:error] [pid 14864:tid 140299653306112] [client 47.145.99.21x:49412] AH10292: Invalid proxy UDS filename (proxy:unix:///var/www/vhosts/system/...
Please sign in to leave a comment.