top button
Flag Notify
    Connect to us
      Site Registration

Site Registration

Mod_proxy Slow After a Week using Httpd

0 votes
460 views

We have:

Apache 2.2.3
CentOS 5.5 x86_64
Splunk 5.0.2

I only know the basics but Apache has been serving us very well with the below config and only after about a week did pages refresh very slowly, up to a minute to reload sometimes.

I tailed both httpd and splunkd logs and saw a consistent delay of 30s between proxy and origin server (both on same host) along with plenty of 304s, followed by what appeared to be some timeout and then slowly the 200s started coming back in.

A restart of httpd cleared up the issue. Im lost, please help. Could this be a caching problem?

posted Jun 24, 2013 by anonymous

Share this question
Facebook Share Button Twitter Share Button LinkedIn Share Button

2 Answers

+1 vote

30 seconds is the length of the default timeout in apache. Unfortunately, that timeout is used in all sorts of cases, so it does not tell us what is timing out.

As a rank guess, I would be going for DNS timeout myself. Do you have HostnameLookups set to "On" or "Double", or using host names in ACLs?

answer Jun 24, 2013 by anonymous
0 votes

You are receiving 304s?That doesnt make sense. Maybe 504s which many times will point to a load issue on the backend server.

answer Jun 24, 2013 by anonymous
Similar Questions
+2 votes

We have a tomcat and apache server on same physical machine. After integration with IDAM, whenever user is accessing the link through IDAM application it is either very slow or not accessible. After Apache restart it start working but after 3-4 hours again same issue occurs.

Tried to do some tuning as follows but no luck.

Timeout 120 
KeepAlive Off 
MaxKeepAliveRequests 100 
KeepAliveTimeout 15 
0 votes

We are currently using HTTP connector in tomcat 7.42 and planning to switch to AJP NIO connector. When I was reading through the docs I found "WARNING: The NIO connector for AJP is experimental."

This made me think that NIO connector might not be mature at this point. Can somebody who had experience with NIO connector let me know if it is stable and any pitfalls I should be aware of?

Stable enough to be used on enterprise apps which is expected to get thousands of requests in a hour?

Any information would be helpful.

+1 vote

We have a Windows XP machine functioning as a file etc "server" and a Debian Linux machine running
Apache httpd as a web server. We intend to replace the XP machine with some version of Windows Server (looking at WHS 2011 at the moment on a trial basis). I would like to be able to continue running the Apache server, but to allow web access to the WHS machine using apache mod_proxy. I already have this working for several servers (etc. usermin).

I've tried setting this up in my mod_proxy ssl config:

ProxyRequests off
SSLProxyEngine on
ProxyPass /whs/ https://x.x.x.x:443/
ProxyPassReverse /whs/ https://x.x.x.x:443/

When I enter https://mydomain.com/whs/ on an external computer I get "Internet Explorer cannot display the webpage". If I try to access this site within my network my browser redirects to
https://x.x.x.x/Remote/logon?ReturnUrl=%2fremote

Has anyone got this config working, or is there a recipe somewhere for this? I really don't want to use IIS as my primary web server.

+1 vote

Im configuring the apache httpd server in front of a farm of cometd servers. The clients interact with the cometd server through long polling connections.
Does the httpd server still use a thread-per-request model? In my usecase this means that a thread per client will be "active". http://cometd.org/node/81

...