Applicable to:
- Plesk for Linux
Symptoms
-
Domain example.com does not work:
PLESK_INFO: 503 service unavailable
PLESK_INFO: 502 Bad Gateway
Nginx -
The following error appears in domain's error log
/var/www/vhosts/system/example.com/logs/error_log:CONFIG_TEXT: [proxy:error] [pid 15065:tid 139826443527936] (111)Connection refused: AH02454: FCGI: attempt to connect to Unix domain socket /data/vhosts/system/example.com/php-fpm.sock (*) failed
[proxy_fcgi:error] [pid 15065:tid 139826443527936] [client 203.0.113.2:50232] AH01079: failed to make connection to backend: httpd-UDS
Cause
Configuration file for a domain is missing in /opt/plesk/php/X.X/etc/php-fpm.d/ directory.
Resolution
Regenerate the configuration:
- Log into Plesk.
- Go to Domains > example.com > PHP Settings and change PHP handler to different one and click OK
- Change PHP handler version back and click OK.
If multiple domains are affected:
-
Connect to the server via SSH;
Note: if direct SSH access to the server is not possible, contact server administrator for further assistance.
-
Execute the following command to update settings on all domains:
# /usr/local/psa/bin/php_settings -u
-
Restart PHP handler processes, for example:
# systemctl restart plesk-php74-fpm
Comments
18 comments
I had this issue and solved it like it is discribed.
The question is now: How did that happen and how to prevent it? As I logged into the plesk panel it showed me that several packages were updated. Could that have been the reason?
Hi @Cornelia Mühlich.
Maybe there was a hung PHP process that was terminated by the actions mentioned in the article.
Possible. But in that case I can't prevent it to happen again... Next time it happens I'll check for that. Thank you (:
First Resolution doesn't work for us (Onyx Plesk 17.5 - Debian 8).
Second Resolution via SSH is a bit an overkill.
When there are many domains on a server this will take a while.
Is there any possibility to use this only for one domain?
@Cfaessler,
Hello! I see that you have already submitted a request to support in regards to your questions.
I believe they will find a solution soon.
Is this related to the issue described in
https://support.plesk.com/hc/en-us/articles/360014718313-Domains-on-Plesk-PHP-7-3-are-down-Service-Unavailable
?
@Cirrus Tecnologia,
Hello! No, the issue described in this article is caused by the corrupted or missing configuration file while the issue from the mentioned article is caused by PHP 7.3 bugs.
Got this same issue the moment after Plesk Migrator finished working. I had one website transfered from one server to another, and that migration caousd this symthoms and it took 30% of my websites down. Any update on making sure that this does not happen?
Hello @spanovic,
Migration is a quite complex process and there are might be various causes of occurred behaviour. Most probably PHP re-configuration process was interrupted due to some reason. However, a thorough investigation anyway is required to define the root cause.
You can submit a ticket for us or use a server-wide workaround from the article.
I just experienced this problem when adjusting modsecurity rules. This is the steps to reproduce
I use the Atomicorp Advanced Ruleset purchased through plesk
PHP-FPM 7.2
I have some rules disabled in a domains Additional Apache Directives (as per this article https://support.plesk.com/hc/en-us/articles/115002531753-How-to-disable-a-single-ModSecurity-rule-for-a-website-in-Plesk
I then tried to disable an extra rule for the same domain by adding it to Modsecurity settings (as per this article https://support.plesk.com/hc/en-us/articles/115002531753-How-to-disable-a-single-ModSecurity-rule-for-a-website-in-Plesk
This caused this error, tested it several times, and it causes this error everytime. To workaround until it is fixed, I have removed the rule I wished to disable from ModSecurity settings, and added it to the list in the Additional Apache Directives.
No issues. I haven't tested to determine if it's having different rules disabled by both methods at the same time, or a fault just with the application of the modsecurity settings
I hope this is helpful
@Christopher Scott
This behavior seems to be unrelated both to the current and the referred articles.
I would suggest contacting regular Plesk support to investigate this behavior precisely.
@...
I have the same symptoms, but the resolution doesn't fix it.
The problem exists since update to Plesk Oblivion.
Do you have any other suggestions?
Ferdinand please open a support ticket with us so we can take a look: https://support.plesk.com/hc/en-us/articles/213608509-How-to-submit-a-request-to-Plesk-support-
Hello,
We have the same problem, some sites were failing to run on PHP-FPM.
The provided fix (switching PHP handler) worked for one addon domain, but failed for a subscription main domain.
For the failed domain:
PHP-FPM socket file is not found.
It works on FastCGI, but problem is back again when reverting to PHP-FPM.
"/usr/local/psa/bin/php_settings -u" is really an overkill when there are a lot of domains.
Can you please provide instructions how to fix the problem manually?
Thank you!
Hello Oleg,
Based on described situation the issue you have should be different, as missing configuration file is rare case.
Please check /var/log/plesk-php7x-fpm/error.log for errors. I might assume you rather hit this issue.
If you sure you have issue with missing configuration file then to fix for single domain it is required to change PHP settings to different version and then back to regenerate configuration.
There was no logs and it was behaving exactly as I described.
After Plesk upgrade to version 18.0.29, the workaround with switching to FastCGI than back to PHP-FPM fixed the problem.
So it looks like being a bug that was fixed.
Thank you!
Oleg Burca
Thank you for letting us know.
The problem has come up again lately
Debian 10.13
Plesk 18.0.49 Update #2
Please sign in to leave a comment.