T
T
tuxx2015-08-25 21:01:21
Apache HTTP Server
tuxx, 2015-08-25 21:01:21

Httpd seems to be hanging?

There is a VPS tariff "a href=" www.smartape.ru/vps "> "Standard" from SmartApe on CentOS 6. It has a site on 1C-Bitrix, which suddenly freezes periodically, i.e. after some random time the server stops loading pages.The only thing that saves is reloading the httpd process or a complete reboot of the VPS.Part of the httpd config
:


ServerTokens OS
ServerRoot "/etc/httpd"
PidFile run/httpd.pid
Timeout 30
KeepAlive Off
MaxKeepAliveRequests 6000
KeepAliveTimeout 10
StartServers 1
MinSpareServers 1
MaxSpareServers 3
ServerLimit 11
MaxClients 10
MaxRequestsPerChild 10000

Such settings were chosen to make the site work more or less stably. Previously, also at random intervals, mysql lost the socket, although at the same time it was running itself and the socket was not used by any other processes. There were also periods when the database was simply killed by oom-killer for exceeding the RAM limit, although usually mysql consumed no more than 300 MB of memory. In both cases, the number of httpd processes was up to several dozen.
And there is still a moment with the fact that there is a constant unloading of the list of changed goods from the 1C database and loading orders from the site. Both rear runs every 10 minutes if the previous runs have already completed. Goods are unloaded in packs of 10 items so as not to load the site too much.
Do not offer a change of tariff, because. the transfer to a more powerful configuration was not easy due to the deafness of the authorities.

Answer the question

In order to leave comments, you need to log in

2 answer(s)
N
neol, 2015-08-25
@neol

Try to remove the launch of Bitrix agents from hits, maybe this will help. If it helps, transfer to crowns.

A
Azazel PW, 2015-08-26
@azazelpw

You yourself wrote
"mysql was losing the socket, although it was running itself"
So you need to look not at the Apache settings, but at the muscle.
Most often, the error occurs in the database server, and not in Apache.
look at the muscle logs. Usually he writes to the log when and why he falls off.
Try changing the settings for memory allocation, the number of connections to the database.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question