Articles in this section

See more

Website running PHP-FPM shows 503 Service Temporarily Unavailable: failed to make connection to backend

Avatar
Stefan Yakubov
Updated
Follow
Plesk for Linux kb: technical ABT: Group A

Applicable to:

  • Plesk for Linux

Symptoms

  • Website is showing the following error:

    CONFIG_TEXT: 503 Service Temporarily Unavailable

  • The following errors are shown in /var/log/plesk/panel.log log file:

    PLESK_INFO: [proxy:error] [pid 12966] AH00959: ap_proxy_connect_backend disabling worker for (127.0.0.1) for 60s
    [proxy_fcgi:error] [pid 12966] [client 203.0.113.2:53377] AH01079: failed to make connection to backend: 127.0.0.1, referer: https://example.com/

  • The website is using PHP-FPM handler at Domains > example.com > PHP Settings

  • The following errors can be found in the website error_log file at /var/www/vhosts/example.com/logs/error_log:

    PLESK_INFO: [proxy:error] [pid 3838:tid 140126940247808] (2)No such file or directory: AH02454: FCGI: attempt to connect to Unix domain socket /var/www/vhosts/system/example.com/php-fpm.sock (*) failed
    [proxy_fcgi:error] [pid 3838:tid 140126940247808] [client 203.0.113.2:56904] AH01079: failed to make connection to backend: httpd-UDS

Cause

PHP-FPM service is stopped.

Resolution

Warning: To apply these steps is required access to the server as Plesk administrator. If you don't have admin access, please contact your service provider.

Click on a section to expand

Via Plesk GUI

  1. Log in to Plesk GUI

  2. Check what version of PHP is used by domain at Domains > example.com > PHP Settings

  3. Go to Tools & Settings > Services Management

  4. Start the PHP-FPM service with the required version. For example the Plesk PHP-FPM 7.3 service:

    mceclip0.png

Via SSH (Single Domain)

  1. Connect to the server via SSH

  2. Get the PHP-FPM version assigned to the domain:

    # plesk db "SELECT d.name,h.php_handler_id FROM domains d JOIN hosting h on h.dom_id=d.id WHERE d.name='example.com'"
    +-------------+----------------+
    | name | php_handler_id |
    +-------------+----------------+
    | example.com | plesk-php73-fpm |
    +-------------+----------------+

  3. Start the PHP-FPM service with the required version. For example the Plesk PHP-FPM 7.3 service:

    # service plesk-php73-fpm restart

Via SSH (Multiple Domains)

  1. Connect to the server via SSH

  2. Restart all the enabled PHP-FPM services:

    # for PHP in `systemctl list-unit-files | grep enabled | grep php.*fpm | awk '{print $1}'`; do echo -e "PHP-FPM service:\t$PHP"; echo -e "Status:\t$(systemctl is-active $PHP)"; echo -e "Restarting..."; systemctl restart $PHP; echo -e "Status:\t$(systemctl is-active $PHP)\n"; done

Additional Information

Domain does not work on PHP-FPM on Plesk: 503 service unavailable

Comments

8 comments

Sort by Date Votes
  • Avatar
    Marco Marsala (Edited )

    Seems that this service is not started at boot. You should start it manually from Tools & Settings > Services Management.

    When I experienced crash of this service today, I restarted the whole server but the service wasn't start at boot.

    In the the log below, for PHP-FPM 5.6 handler, you can see that service never tried to start at boot. Reboot was done after 08:46, time of last status log entry for such service. Starting service manually worked without any issue. The same applies to PHP-FPM 7.1 handler.

    This server is Ubuntu 16.04.3 LTS (upgraded from 14.04 following the KB procedure) with Plesk 17.5.3 Update #28.

    # /etc/init.d/plesk-php56-fpm status
    ● plesk-php56-fpm.service - The PHP 5.6.31 FastCGI Process Manager
     Loaded: loaded (/lib/systemd/system/plesk-php56-fpm.service; enabled; vendor preset: enabled)
     Active: failed (Result: signal) since Thu 2017-11-09 08:46:52 CET; 16min ago
     Process: 1119 ExecStart=/opt/plesk/php/5.6/sbin/php-fpm --nodaemonize (code=killed, signal=KILL)
     Main PID: 1119 (code=killed, signal=KILL)
     Status: "Processes active: 5, idle: 0, Requests: 0, slow: 0, Traffic: 0req/sec"
     CGroup: /system.slice/plesk-php56-fpm.service

    Nov 09 08:43:52 thenetworksolution.it systemd[1]: Starting The PHP 5.6.31 FastCGI Process Manager...
    Nov 09 08:45:22 thenetworksolution.it systemd[1]: plesk-php56-fpm.service: Start operation timed out. Terminating.
    Nov 09 08:46:52 thenetworksolution.it systemd[1]: plesk-php56-fpm.service: State 'stop-final-sigterm' timed out. Killing.
    Nov 09 08:46:52 thenetworksolution.it systemd[1]: plesk-php56-fpm.service: Main process exited, code=killed, status=9/KILL
    Nov 09 08:46:52 thenetworksolution.it systemd[1]: Failed to start The PHP 5.6.31 FastCGI Process Manager.
    Nov 09 08:46:52 thenetworksolution.it systemd[1]: plesk-php56-fpm.service: Unit entered failed state.
    Nov 09 08:46:52 thenetworksolution.it systemd[1]: plesk-php56-fpm.service: Failed with result 'signal'.
    0
    Comment actions Permalink
  • Avatar
    Denis Bykov (Edited )

    @Marco
    As I can see, the service is enabled, and the only explanation why it won't start at the boot is a failed dependency. plesk-php56-fpm depend on network and syslog:

    # grep After /lib/systemd/system/plesk-php56-fpm.service
    After=syslog.target network.target

    Try to check if syslog works correctly:

    # systemctl status syslog.service
    1
    Comment actions Permalink
  • Avatar
    Marco Marsala

    @Denis dependencies are ok, otherwise launching the service manually will fail too. As I said, starting the service manually worked.

    Btw I checked the syslog:

    ~# systemctl status syslog.service
    ● rsyslog.service - System Logging Service
     Loaded: loaded (/lib/systemd/system/rsyslog.service; enabled; vendor preset: enabled)
     Active: active (running) since Thu 2017-11-09 08:43:53 CET; 6 days ago
     Docs: man:rsyslogd(8)
     http://www.rsyslog.com/doc/
     Main PID: 920 (rsyslogd)
     CGroup: /system.slice/rsyslog.service
     └─920 /usr/sbin/rsyslogd -n

     

    0
    Comment actions Permalink
  • Avatar
    Denis Bykov

    @Marco

    Are there some related errors in '/var/log/plesk-php56-fpm/error.log' and '/var/log/plesk-php71-fpm/error.log' files?

    If not, and the issue can be reproduced, I suggest raising a support ticket regarding this behavior.

    0
    Comment actions Permalink
  • Avatar
    Florian

    Encountering this error when trying to access .php files, any ideas?

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    @Florian

    Hello!

    Such behavior may be caused by PHP-FPM service. Try to apply this article.

    In case the issue will persist, consider contacting Plesk Support.

    0
    Comment actions Permalink
  • Avatar
    Marco Marsala (Edited )

    @Denis There isn’t any error in such log files, because the service is not called at all. Seems this happens after performing Ubuntu 14>16 upgrade, following KB instructions, that broke some things. Maybe the installer doesn’t reconfigure some services to start at boot. This happened with other services and it was classified as bug.

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    @Marco Marsala

    Hi!

    As @Denis Bykov mentioned, we suggest raising a support ticket regarding this behavior.

    0
    Comment actions Permalink

Please sign in to leave a comment.

Have more questions? Submit a request

Related articles