Resolved Package Updater says failure when OS-PHP-FPM pool is empty

[TimReeves]

I'm running Debian 8 and Plesk Onyx 17.5.3 #37.
I have noticed this problem on more than one machine: The OS-PHP FPM pool is empty, since I now only use PHP7.0 and 7.1. When I do a manual call of apt-get upgrade the output looks like this:

Setting up php5-fpm (5.6.33+dfsg-0+deb8u1) ...
php5_invoke mysql: no action - module was disabled by maintainer for fpm SAPI
php5_invoke imap: no action - module was disabled by maintainer for fpm SAPI
php5_invoke sqlite3: no action - module was disabled by maintainer for fpm SAPI
php5_invoke pdo_sqlite: no action - module was disabled by maintainer for fpm SAPI
Job for php5-fpm.service failed. See 'systemctl status php5-fpm.service' and 'journalctl -xn' for details.
invoke-rc.d: initscript php5-fpm, action "start" failed.
dpkg: error processing package php5-fpm (--configure):
 subprocess installed post-installation script returned error exit status 1
Setting up linux-libc-dev:amd64 (3.16.51-3+deb8u1) ...
Errors were encountered while processing:
 php5-fpm
E: Sub-process /usr/bin/dpkg returned an error code (1)

This results in an annoying E-Mail from Plesk like this:

Reason: 2018-01-15 06:33:02 INFO: pum is called with arguments: ['--update', '--json']
2018-01-15 06:33:21 INFO: updating packages: python-libxml2 libxml2 libxml2-dev
Error in function:
SystemError: E:Sub-process /usr/bin/dpkg returned an error code (1)
2018-01-15 06:33:29 ERROR: installArchives() failed
2018-01-15 06:33:29 ERROR: Exited with returncode 1.

I don't know if I could really delete the OS PHP package, I don't know if other things rely on it, like Plesk itself, or Apache, or anything else. And unless I do, I suspect that I will keep getting these misleading E-Mails - unless I configure a dummy PHP-FPM process with nothing to do... ahem.

Just one idea: If the Plesk package update manager could scan the output from apt, and if it is ONLY complaining about PHP for an empty pool not starting, then discount the error message.

 

[IgorG]

Try to fix it with:

# plesk installer --select-release-current --reinstall-patch --upgrade-installed-components
# plesk sbin pum --update

 

[TimReeves]

Hi Igor,

thanks for that idea, I applied it yesterday at 2 servers with this problem. This morning I did not see that problem.

Thanks!

 

[jmar83]

I had a similar problem with Plesk Obsidian 18.0.28 Update #3... which occured after a failed update... all usernames from the domains saved in directory /etc/php5/fpm/pool.d/... didn't match anymore with the real / correct usernames in plesk web GUI...

Then i tried 1.) to change all hosts to php 7.x; after that `apt-get purge php5-fpm` and install it again with `apt-get install php5-fpm`... but that didn't deleted the content with non-equivalent usernames from /etc/php5/fpm/pool.d/... , so `apt-get install php5-fpm` tried to start php5-fpm with non-existing usernames after installation - which resulted with an error code not equals 0 2.) then i moved the whole content from /etc/php5/fpm/pool.d/... to a new created directory in root: /pool.d.. after that' i tried to uninstall it again, then occured another error message (i forgot it)... after deleting the whole /etc/php5/fpm directory, i was successful with purge & (re)-install php5-fpm...

Then a "apt-get clean && apt-get autoremove --purge" and the system seems to be "clean" again.

I still don't know what the problem was with that non-equivalent usernames. (???) but anyway, now it works again.

 

[jmar83]

Was that an adequate way to solve the problem, or do i have an inconsistent system now?

Thank you for your feedbacks.


Best regards, Jan

 

[jmar83]

Maybe that question is too complex for that forum??

 

[jmar83]

Seems so...!!?