502 Bad Gateway Errors with HostBill

Started by BRJP, August 03, 2014, 02:58:16 PM

BRJP

Dear All,

This is something that has plagued us for a while but we never really seem to solve it.

We randomly receive Gateway Errors (see attached screenshot) when using HostBill.  The NGINX log shows us the below error which is typical when this occurs.

2014/08/03 19:42:05 [error] 26027#0: *11549 upstream prematurely closed connection while reading response header from upstream, client: 89.x44.2x4.xxx, server: cloned.com, request: "GET /myportal/?/checkdomain/domain-registration/ HTTP/1.1", upstream: "https://212.x7.2xx.x0:7081/myportal/?/checkdomain/domain-registration/", host: "www.cloned.com", referrer: "http://www.cloned.com/support.html"

All our websites on the server never receive this problem.  Our main website never has this issue.  HostBill is within a subdirectory of our main website and we keep coming across this error - so it does seem like something specific with HostBill either causing or aggravating an issue.

Any thoughts on how we look deeper into this as our techie and Google don't seem to come across any solutions that seem viable or work!

Thanks in advance.

BRJP
Kind regards,
Bradley Porter
--------------------
Find out more about SaneChoice Services at: https://www.sanechoice.cloud/

AmazingAlexx

#1
Increase your buffer size, should fix the issue

BRJP

Quote from: AmazingAlexx on August 03, 2014, 03:49:27 PM
Increase your buffer size, should fix the issue

Thank you.  The buffer size of what?
Kind regards,
Bradley Porter
--------------------
Find out more about SaneChoice Services at: https://www.sanechoice.cloud/

AmazingAlexx


BRJP

Thank you for the response.  Much appreciated.

We had seem most of them before but the below suggestion is what caught our eye:

Disable APC Cache if used, instead try Xcache, apc can cause this kind of issues under particular enviroments causing segmentation faults.

We had seen segmentation faults as well so this new revelation will be placed into test and hammered.

I will post an update if this indeed fixes this ongoing issue.
Kind regards,
Bradley Porter
--------------------
Find out more about SaneChoice Services at: https://www.sanechoice.cloud/

BRJP

We have solved this issue by using PHP-FPM support for NGINX which is now included in PLESK.  The response time of the website is now very much faster and we do not get the 500 errors.  (Although we are still monitoring with some customer who have complained).
Kind regards,
Bradley Porter
--------------------
Find out more about SaneChoice Services at: https://www.sanechoice.cloud/