Plesk 11.5 - 502 bad gateway

[Dree]

Any Plesk people here who can get this checked out?
I'd rather not open a support ticket and go through a painful 5-day wait, re-explaining the problem 5 times, but perhaps I should anyway

 

[Dree]

Quick fix from support this time, nice! Its working fine now.


1)Take backup of "/etc/sw-cp-server/config"
2)Increase the following buffer sizes in the file "/etc/sw-cp-server/config".

a) fastcgibuffers 32 32k;
b) fastcgibuffer_size 64k;

After editing the file, restart the below services:

/etc/init.d/sw-cp-server restart
/etc/init.d/sw-engine restart

 

[VinnyT]

I try to restart apache, psa and ngix, but not work.
After reboot entire server, works.

I am having this same issue. I will be responding with more detail later, but in the meantime, the way to quickly fix it is to restart the PHP-FPM service "service php-fpm restart"

 

[Silvester van der Leer]

I am having this same issue. I will be responding with more detail later, but in the meantime, the way to quickly fix it is to restart the PHP-FPM service "service php-fpm restart"

Have you tried the fix above your post ? it works for me

 

[VinnyT]

I saw it after I posted your reply. I actually just tried it. Will wait and see. Thank you.

 

[ManfredWarta]

Same Problem on key activation since plesk cp running on nginx:

*1 upstream timed out (110: Connection timed out) while reading response header from upstream, client: xxx.xxx.xxx.xxx, server: , request: "GET /server/key_update.php?key_type=plesk HTTP/1.1", upstream: "fastcgi://unix:/var/run/sw-engine.sock", host: "xxxxx.xxxxxxxx.xx:8443", referrer: "https://xxxxx.xxxxxxxx.xx:8443/server/key_info.php"

 

[VinnyT]

Quick fix from support this time, nice! Its working fine now.


1)Take backup of "/etc/sw-cp-server/config"
2)Increase the following buffer sizes in the file "/etc/sw-cp-server/config".

a) fastcgibuffers 32 32k;
b) fastcgibuffer_size 64k;

After editing the file, restart the below services:

/etc/init.d/sw-cp-server restart
/etc/init.d/sw-engine restart

Update -- changing these values has helped, but it does not fix the issue. An important item to note is that I have about 30 domains on my server, each using php-fpm. When the errors start occurring, it only happens for one of the domains (and the same domain every time). Even though the other domains continue to work, i still have to restart PHP-FPM in order to fix the domain that has the error. It's almost as if PHP-FPM is not properly dropping old processes (keepalive?) and they eventually fill up the total amount of allowed pHP-FPM processes for that single domain. Any ideas?

 

[IgorG]

Have you seen this issue after installation of MU#8?

 

[netsetter]

It seems to be fixed now, thank you.

 

[VinnyT]

Have you seen this issue after installation of MU#8?

Hi Igor -- Everythiing was looking good for a few days, but the error just happened again. 502 - Bad Gateway - NGINX --
Again, this only happens with a SINGLE domain. The rest continue to work. Are there any logs you think I can check to start to identify the issue?

 

[VinnyT]

A quick note -- i looked in the php-fpm logs and noticed the following: "server reached max_children setting (5), consider raising it"
So i went into the /etc/php-fpm.d/ folder and opened the specific domain .conf file.
The max_children variable was set to 5. I changed this value to 25.

An interesting item to note is that the generic www.conf file has the "max_children" value set to 50. Should this number be the same across all domains? Perhaps 5 is way too low for high traffic sites..

 

[VinnyT]

Still no luck for me.

 

[AnorionilB]

Still no luck for me.

Same here. I have the same issue. Tried rebooting the server, this worked and I could run 1 export from phpmyadmin, but then the same error again. I can't keep rebooting the server. Is there any other way to fix this?
php-fpm is an unrecognised service on my server. And the values config in sw-cp-server were already the values described in the post above...

Any other solutions?

 

[IgorG]

MU#14 - http://kb.parallels.com/117127

[-] (Linux only) PhpMyAdmin failed to export databases with the error "502 Bad Gateway" (141734)

 

[AnorionilB]

MU#14 - http://kb.parallels.com/117127

Thank you, this fixed it

 

[SalvadorS]

Hello,

Still same problem here after MU 14. Do I have to do anything else after MU14 to fix the problem?

I restart sw-cp-server and sw-engine but it does not work...

This is the error in /var/log/sw-cp-server/error_log


2013/09/17 18:08:54 [error] 31400#0: *80 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 188.77.xxx.xxx, server: , request: "POST /domains/databases/phpMyAdmin/export.php HTTP/1.1", upstream: "fastcgi://unix:/var/run/sw-engine.sock:", host: "XXX.net:8443", referrer: "https://XXX.net:8443/domains/databases/phpMyAdmin/tbl_export.php?db=XXX_DESARROLLO&table=products_groups&token=727b672a47b40f720ed0bbf3b6cb59b5&single_table=true"

Regards

 

[Silvester Leer]

MU#14 - http://kb.parallels.com/117127

No it's not fixed !

 

[VinnyT]

I haven't had any problems in 2 days since the update... so far so good.

 

[DoniyorS]

Previvously i manually edited sw-cp-server/config file and resolved the problem, but now after update i had it back.

 

[StéphanS]

Just do

# service psa restart
Stopping sw-engine-fpm:                                    [  OK  ]
Starting sw-engine-fpm:                                    [  OK  ]

This might fix it for you (it did for me).