Website running on Plesk with PHP-FPM handler shows error: 503 Service Temporarily Unavailable: php-fpm.sock: No such file or directory

Follow

Comments

14 comments

  • Avatar
    Horacio Stolovitzky

    I'm experiencing this problem frequently in WordPress sites.  Now I got this error in any site and any server.

    Any news for a new release that fix this error?

     

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    @Horacio Stolovitzky

    Hello!

    Plesk developers are doing their best to resolve this issue.
    However, there is no ETA for the resolution.

    0
    Comment actions Permalink
  • Avatar
    Gabriele Roveda

    Good morning,
    I have the same problem with a newly installed server.
    Changing to FAST-CGI instead of FPM you can install and run Joomla, but not Worldpress. With Worldpress there are problems of all kinds. Whether you install from the application or use Worldpress Kit. By mistake on the site path. From error in showing the interface. If you enter the link to log in as wp-admin does not find the page. A disaster! It only works if I load an existing site and its database and not all sites. This happens whether you are keeping the old version of Plesk or updating it. Even if I set PHP 5.6 or all subsequent releases. A real disaster. The paradox is that a week ago we installed another server always with Plesk and everything works there. : ((
    If you do not find a solution quickly we will be forced to disable Plesk and change management app. Too bad because it is your loyal users, but it is impossible to work.

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    Hello @Gabriele,

    Thank you for the feedback.

    I have verified that the bug PPPM-5819 was already resolved in the upcoming Plesk release. Currently, there is no exact release date but as soon as the fix will be publicly released, the article will be updated. Click "Follow" button at this page to be notified.

    Until fix will be available, I can only suggest using the workaround from the "resolution" section.

    In case issue symptoms on your server anyhow differ from symptoms in this article, I would suggest submitting a request for technical support.

    0
    Comment actions Permalink
  • Avatar
    Gabriele Roveda

    Hello Ivan, the problem is also present in the latest release of Plesk Onix. At the moment we are trying to solve it with the assistance of the Aruba partner. If this can be useful I can tell you that just a week ago we installed another server with Centos 7 and Plesk 17.5.3 without updating Plesk to 17.8.11 and everything works perfectly. This new server with the same settings instead no. Almost certainly it is a bug of Centos. Something I found in the forum, but all the proposed solutions do not work. Now I'm waiting for Aruba's assistance to intervene on the Centos Engine. Maybe so the problem is solved. However, this affects the use of Plesk. Especially with Wordpress.

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    Hello @Gabriele,

    Thank you for clarification and additional information.

    > At the moment we are trying to solve it with the assistance of the Aruba partner.

    As the issue is rather complicated and requires detailed investigation on the server this is one of the best options. Our partners are trained experts in Plesk. In case they will have difficulties, they will contact Plesk Technical support for further assistance.

    Also being Plesk partner's client you may contact Plesk Technical Support directly with a support subscription.

    0
    Comment actions Permalink
  • Avatar
    Gabriele Roveda

    In fact they have solved it, there is a bug in Centos, in the Engine and partly also in Plesk in ngix. The problem in conclusion is derived from an incongruity with the ngnix services that was solved by reinstallation and rehabilitation of the ngnix service itself. In fact you can not connect the loading of the interfaces of the domains created in Wordpress and Joomla. We have not tried any more apps yet, but I think Magento will surely have problems too. The fact is that this way, every time you upgrade Plesk you have to pray that nothing happens, and since we do not use it for work and not for hobbies, you understand that this is a serious problem that makes us waste a lot of time and as a result, money. My advice is to do all the possible tests before releasing a new version. Because if it is true that you release the update and then you discover the bugs, it is also true that before issuing the update all the possible tests must be done.

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    Hello @Gabriele,

    Glad to know it was solved on this server.

    Thank you for the valuable feedback. It will be delivered to the corresponding department.

    The article was edited, the information that the bug fix will present in the next major release is present now. 

    0
    Comment actions Permalink
  • Avatar
    Gabriele Roveda

    Happy to be useful. However, please do the debugging tests before releasing updates. Otherwise problems will always arise. Plek is not an open source and if you have to pay it is important that it is managed professionally. Thank you.

    0
    Comment actions Permalink
  • Avatar
    Ekaterina Babenko

    Hello @Gabriele,

    We are always doing our best to improve product functionality and putting all our efforts to make it better. Many thanks for your attention!

    0
    Comment actions Permalink
  • Avatar
    Tony

    Has this bug been fixed in Plesk Version 17.8.11 Update #49, last updated on April 9, 2019?  It was resolved by switching php versions in plesk for the domain as suggested above.

    I had this error in my logs from overnight.

    ```
    [Fri Apr 12 07:51:54.623182 2019] [proxy_fcgi:error] [pid 55486] [client 192.168.0.1:51190] AH01079: failed to make connection to backend: httpd-UDS, referer: https://example.com/
    [Fri Apr 12 07:51:54.696329 2019] [proxy:error] [pid 142340] (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
    ```

    0
    Comment actions Permalink
  • Avatar
    Ivan Postnikov

    Hello @Tony,

    This Plesk version does not contain this fix.

    As stated in the article, the fix will be available in next major release (currently, no exact ETA for this) also fix is already available in Plesk 17.9 (please note that this is a preview version not for production usage).

    After the fix will be publicly available, the article will be updated.

    1
    Comment actions Permalink
  • Avatar
    burnleyvic

    We definitely need this fix backported. If not in 17.5, at least in 17.8, can't afford to wait for the next major release to become production ready, then schedule outage windows for the upgrade and so on and so forth.

    0
    Comment actions Permalink
  • Avatar
    Nikita Nikushkin

    Hi @Adi Pircalabu,

    Fix for this bug requires big changes in the business logic of the product

    Thus, the bug was fixed in the next major Plesk version only where the business logic was reworked

    As for now, the only way to resolve the issue for the supported Plesk versions is switching PHP Handler for the affected domain to another PHP-FPM handler and back to initial one. In case when many domains are affected, the provided script can be used

    0
    Comment actions Permalink

Please sign in to leave a comment.

Have more questions? Submit a request