Apache/FPM crashing from time to time

[ehrenwert]

Hello,

on one of our servers we got the following problem:

Some webpages are showing the server error "Service Unavailable" from time to time.

The logs are showing the following:

[Wed Dec 16 20:44:12.076194 2015] [proxy:error] [pid 23891:tid 140289989596928] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:13.084705 2015] [proxy:error] [pid 23891:tid 140290115487488] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:31.123884 2015] [proxy:error] [pid 23891:tid 140289981204224] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:31.632863 2015] [proxy:error] [pid 23891:tid 140290090309376] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:31.983987 2015] [proxy:error] [pid 23891:tid 140290014775040] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:32.113282 2015] [proxy:error] [pid 23891:tid 140290247546624] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:45.637922 2015] [proxy:error] [pid 23891:tid 140290023167744] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:44:46.348945 2015] [proxy:error] [pid 23891:tid 140290081916672] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:53:45.563478 2015] [proxy:error] [pid 23891:tid 140290023167744] (111)Connection refused: AH02454: FCGI: attempt to connect to Unix domain socket /var/www/vhosts/sys$

[Wed Dec 16 19:53:45.563496 2015] [proxy:error] [pid 23891:tid 140290023167744] AH00959: ap_proxy_connect_backend disabling worker for (127.0.0.1) for 60s

[Wed Dec 16 19:53:45.563499 2015] [proxy_fcgi:error] [pid 23891:tid 140290023167744] [client 93.130.41.9:54220] AH01079: failed to make connection to backend: httpd-UDS, referer: h$

[Wed Dec 16 19:53:52.838754 2015] [proxy:error] [pid 23891:tid 140290222343936] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:54:02.313917 2015] [proxy:error] [pid 23891:tid 140290031560448] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:54:12.467022 2015] [proxy:error] [pid 23891:tid 140290230744832] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:55:21.002539 2015] [proxy:error] [pid 23891:tid 140290230744832] (111)Connection refused: AH02454: FCGI: attempt to connect to Unix domain socket /var/www/vhosts/sys$

[Wed Dec 16 19:55:21.002563 2015] [proxy:error] [pid 23891:tid 140290230744832] AH00959: ap_proxy_connect_backend disabling worker for (127.0.0.1) for 60s

[Wed Dec 16 19:55:21.002573 2015] [proxy_fcgi:error] [pid 23891:tid 140290230744832] [client 192.241.129.131:56642] AH01079: failed to make connection to backend: httpd-UDS

[Wed Dec 16 19:55:47.161574 2015] [proxy:error] [pid 23891:tid 140290107094784] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:55:47.824909 2015] [proxy:error] [pid 23891:tid 140290090309376] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:55:47.963637 2015] [proxy:error] [pid 23891:tid 140290073523968] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:56:07.404914 2015] [proxy:error] [pid 23891:tid 140290031560448] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 19:59:59.515347 2015] [proxy:error] [pid 23891:tid 140290014775040] (111)Connection refused: AH02454: FCGI: attempt to connect to Unix domain socket /var/www/vhosts/sys$

[Wed Dec 16 19:59:59.515366 2015] [proxy:error] [pid 23891:tid 140290014775040] AH00959: ap_proxy_connect_backend disabling worker for (127.0.0.1) for 60s

[Wed Dec 16 19:59:59.515370 2015] [proxy_fcgi:error] [pid 23891:tid 140290014775040] [client 77.119.129.165:37826] AH01079: failed to make connection to backend: httpd-UDS, referer$

[Wed Dec 16 19:59:59.752275 2015] [proxy:error] [pid 23891:tid 140290006382336] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:00:21.011796 2015] [proxy:error] [pid 23891:tid 140290115487488] AH00940: FCGI: disabled connection for (127.0.0.1)

[Wed Dec 16 20:01:21.246016 2015] [proxy:error] [pid 23891:tid 140290247546624] (111)Connection refused: AH02454: FCGI: attempt to connect to Unix domain socket /var/www/vhosts/sys$

[Wed Dec 16 20:01:21.246041 2015] [proxy:error] [pid 23891:tid 140290247546624] AH00959: ap_proxy_connect_backend disabling worker for (127.0.0.1) for 60s

[Wed Dec 16 20:01:21.246045 2015] [proxy_fcgi:error] [pid 23891:tid 140290247546624] [client 91.113.89.69:39282] AH01079: failed to make connection to backend: httpd-UDS, referer: $

The PHP5-FPM.log shows the following:

[16-Dec-2015 19:49:12] WARNING: [pool example.com] server reached max_children setting (5), consider raising it

The server has a lot of free resources and is not under DDoS/Brute Force.

Any ideas?

 

[IgorG]

Try to apply solution from KB article https://kb.odin.com/en/119314

 

[ehrenwert]

I did already do that - without any success:
/etc/apache/conf.d/swtune.conf

<IfModule worker.c>

# KeepAlive On

# MaxKeepAliveRequests 10

# KeepAliveTimeout 2



StartServers 10

MinSpareServers 50

MaxSpareServers 150

ServerLimit 1550

MaxClients 3000

MaxRequestsPerChild 4100

MaxRequestWorkers 500

</IfModule>



<IfModule prefork.c>

# KeepAlive On

# MaxKeepAliveRequests 10

# KeepAliveTimeout 2



StartServers 10

MinSpareServers 50

MaxSpareServers 150

ServerLimit 1550

MaxClients 3000

MaxRequestsPerChild 4100

MaxRequestWorkers 500

</IfModule>



<IfModule mpm_prefork_module>

# KeepAlive On

# MaxKeepAliveRequests 10

# KeepAliveTimeout 2



StartServers 10

MinSpareServers 50

MaxSpareServers 150

ServerLimit 1550

MaxClients 3000

MaxRequestsPerChild 4100

MaxRequestWorkers 500

</IfModule>

 

[IgorG]

Is this Plesk installed on VPS or standalone server? If it is Plesk 12.5 version - have you tried

# plesk repair web

?

 

[ehrenwert]

It happens on one VPS (powerful - much free resources) and on a dedicated server aswell.

What does "plesk repair web" actually do? Does it cause downtime? I had to edit some configuration files - are they overwritten by that?

 

[IgorG]

What does "plesk repair web" actually do? Does it cause downtime? I had to edit some configuration files - are they overwritten by that?

http://docs.plesk.com/en-US/12.5/ad...epair-utility/plesk-repair-utility-web.74654/