Question Install plesk php 7.3 kills other running php in other subscriptions

[Azurel]

I installed plesk-php version 7.3 get status 503/502 in other subscriptions with running plesk-php 7.2 and its broke user uploads and more things. Is here a way to install without breaking/kill services?

 

[Monty]

I've installed PHP 7.3 via Plesk installer on over 20 servers without any issues, so I don't think it's a general issue.

What does the error.log of your plesk-php72-fpm say?

 

[Azurel]

/var/log/plesk/install/ai_action_time.xml

<action version="1.0.0" timestamp="20190320T181653" duration="61" status="success">
<release version="PLESK_17_8_11" mode="update" />
<components>
<component name="plesk/panel" mode="U" from="14081815-centos7.17100418.9e86c2f" to="14081815-centos7.17100418.9e86c2f" />
<component name="plesk/modsecurity" mode="U" from="1:2.9.1-33.el7.art" to="1:2.9.1-33.el7.art" />
<component name="plesk/php7.3" mode="I" />
</components>
</action>

/var/log/plesk-php73-fpm/error.log

[20-Mar-2019 18:17:45] NOTICE: configuration file /opt/plesk/php/7.3/etc/php-fpm.conf test is successful

[20-Mar-2019 18:17:47] NOTICE: fpm is running, pid 12441
[20-Mar-2019 18:17:47] NOTICE: ready to handle connections
[20-Mar-2019 18:17:47] NOTICE: systemd monitor interval set to 10000ms
[20-Mar-2019 18:18:06] NOTICE: configuration file /opt/plesk/php/7.3/etc/php-fpm.conf test is successful

[20-Mar-2019 18:18:06] NOTICE: Reloading in progress ...
[20-Mar-2019 18:18:06] NOTICE: reloading: execvp("/opt/plesk/php/7.3/sbin/php-fpm", {"/opt/plesk/php/7.3/sbin/php-fpm", "--nodaemonize"})

/var/log/plesk-php72-fpm/error.log

[20-Mar-2019 18:17:44] NOTICE: fpm is running, pid 7637
[20-Mar-2019 18:17:44] NOTICE: ready to handle connections
[20-Mar-2019 18:17:44] NOTICE: systemd monitor interval set to 10000ms
[20-Mar-2019 18:17:45] NOTICE: configuration file /opt/plesk/php/7.2/etc/php-fpm.conf test is successful

[20-Mar-2019 18:17:46] NOTICE: configuration file /opt/plesk/php/7.2/etc/php-fpm.conf test is successful

[20-Mar-2019 18:17:46] NOTICE: configuration file /opt/plesk/php/7.2/etc/php-fpm.conf test is successful

[20-Mar-2019 18:17:48] NOTICE: Reloading in progress ...
[20-Mar-2019 18:17:48] ERROR: unable to read what child say: Bad file descriptor (9)
[20-Mar-2019 18:17:48] ERROR: unable to read what child say: Bad file descriptor (9)
[20-Mar-2019 18:17:48] NOTICE: reloading: execvp("/opt/plesk/php/7.2/sbin/php-fpm", {"/opt/plesk/php/7.2/sbin/php-fpm", "--nodaemonize"})

I see that in /vhost/system/example.com/conf/nginx.conf was changed by a user(GUI) in time 18:20:56 but find nothing in error.log. Is this not part of it?

 

[Monty]

I assume restarting plesk-php72-fpm solved your problem, right?

The "unable to read what child say: Bad file descriptor (9)" error can have a variety of reasons. If it only happened once then consider it a case of "bad luck".

Additionally, check those threads/articles for further hints:
Issue - PHP70-FPM ERROR: unable to read what child say: Bad file descriptor (9)
Websites hosted in Plesk and served by PHP-FPM are unavailable or load slowly: "server reached max_children setting" or "Bad file descriptor (9)"

 

[Azurel]

Thanks for bad file descriptor links. I will try value 20 for "pm.max_children".