[PHP 503 error] AH01079: failed to make connection to backend: httpd-UDS

kastin

New member
Joined
Jan 5, 2015
Messages
2
DirectAdmin Values
Compiled on CentOS 7.0 64-Bit
Server Version 1.46.3

PHP: php-fpm 5.6

Fresh install DA, .php files not work - 503 error. At /var/log/httpd/domain.error.log see:
Code:
# AH01079: failed to make connection to backend: httpd-UDS

Solution:
http://forum.directadmin.com/showthread.php?t=49969
Code:
# systemctl restart php-fpm56.service

imo. it should be executed during installation
 
I hit the same issue on centos7, this should be part of the install routines etc
 
Hmm.. DA should start them up when it notices they're not running, which is how most services get their initial start-up.
Check:
Code:
cat /usr/local/directadmin/data/admin/services.status
to ensure you see php-fpm56=ON listed.
If not, let me now and we'd have to check why it's not present.

John
 
Hmm.. DA should start them up when it notices they're not running, which is how most services get their initial start-up.
Check:
Code:
cat /usr/local/directadmin/data/admin/services.status
to ensure you see php-fpm56=ON listed.
If not, let me now and we'd have to check why it's not present.

John

Confirmed, the value is present but service is not started. Manually starting resolves the issues.

Tested with:
CentOS 7.0 - 3.10.0-514.el7.x86_64
Directadmin Custombuild 2.0.0 (rev: 1663)
 
Did you solve the issue?

We have this issue with DA apache/nginx combo, php-fpm and php 7.0 and php 5.6 installed at CentOS Linux release 7.3.1611. I downgraded today from 7.1 to 7.0 because of Magento 2. But now I am having 503 errors every couple of minutes. Services restart automatically, but it shouldn't go down anyway.

I reinstalled apache_nginx combo and php today, but that didn't make a difference.
 
Last edited:
Nginx is also terminating itself, no clue why?

Code:
[01-Jun-2017 17:07:07] NOTICE: Terminating ...
[01-Jun-2017 17:07:07] NOTICE: exiting, bye-bye!
[01-Jun-2017 17:08:01] NOTICE: fpm is running, pid 13600
[01-Jun-2017 17:08:01] NOTICE: ready to handle connections
[01-Jun-2017 17:09:31] NOTICE: Terminating ...
[01-Jun-2017 17:09:31] NOTICE: exiting, bye-bye!
[01-Jun-2017 17:09:36] NOTICE: fpm is running, pid 13650
[01-Jun-2017 17:09:36] NOTICE: ready to handle connections
 
Production server?

We had this kind last year, centos 7 php 7 and php 5.6 fpm only apache.

Reboot server then the problem exists ( seemed to start but...), so didn't know wich update before or changes, also not when was causing this.

Deinstall php 7 didn't work same probs

Time saving for us was placing back snapshot image and then restore backups. ( because mostly al rebuilds/ing reinstall rewrite confs and so on didn't work to)

Sorry i hope you or someone else find the solution for these kind of probs.

support ( hosting company) didn't solved it while we do not have a "default" ( that time http2, all fpm, mariadb and letsencrypt and some minor things) install they broke more, so....

Take a look at all with fpm, and hard reboot. (my thoughts then something in combi with fpm...)

cache progs ... proxy's

security things protecting deinstallations/changes partly...

ssl changes php / apache nginx against the ssl from centOS


For Forum readers information after that time we only use one php version on DA boxes!
( sometimes httpd is down (a restart was going wrong) but still one/some httpd running in the back, few times a year, reboot is working then) Support did find out there was still a running httpd in the background.

killall -9 httpd
service httpd start
is not working then kill do not find..

i did write this here in forum
http://forum.directadmin.com/showthread.php?t=53999&p=276902#post276902
 
Last edited:
Still having this problem, with php 7.0 only. Not with 7.1 or 5.6.

I found this other topic where someone refers to a missing folder. https://support.plesk.com/hc/en-us/articles/213959325-PHP-7-0-FPM-does-not-work-for-domains

I changed log_level to "debug" at /usr/local/php70/etc/php-fpm.conf

No errors at
Code:
/var/log/php-fpm70.log
but a few notices that could lead to the bug:

Code:
[09-Jun-2017 09:00:44.563247] NOTICE: pid 1204, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:02:40.602659] NOTICE: pid 1059, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:04:32.102164] NOTICE: pid 2403, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:06:07.353434] NOTICE: pid 7421, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:08:31.601210] NOTICE: pid 7638, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:10:06.852145] NOTICE: pid 7682, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:12:32.104692] NOTICE: pid 7750, fpm_pctl_exit(), line 70: exiting, bye-bye!
[09-Jun-2017 09:14:07.350115] NOTICE: pid 7803, fpm_pctl_exit(), line 70: exiting, bye-bye!

and

Code:
[09-Jun-2017 09:42:07.101706] DEBUG: pid 9183, fpm_got_signal(), line 85: received SIGTERM
[09-Jun-2017 09:42:07.101761] NOTICE: pid 9183, fpm_got_signal(), line 86: Terminating ...
[09-Jun-2017 09:42:07.101766] DEBUG: pid 9183, fpm_pctl(), line 231: switching to 'terminating' state
[09-Jun-2017 09:42:07.101769] NOTICE: pid 9183, fpm_pctl_exit(), line 70: exiting, bye-bye!

and once and a while:
Code:
[09-Jun-2017 10:01:18.081776] DEBUG: pid 11320, fpm_children_make(), line 421: [pool misss_useraccount] child 11396 started
[09-Jun-2017 10:01:18.081815] DEBUG: pid 11320, fpm_pctl_on_socket_accept(), line 536: [pool misss_useraccount] got accept without idle child available .... I forked
 
Last edited:
Back
Top