httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kelsey, Paul" <PKEL...@indygov.org>
Subject [users@httpd] httpd hangs until Max Clients is reached
Date Mon, 11 Aug 2008 17:16:09 GMT
Hello All,

 

I have a Sun/Solaris 8 system runing Apache 1.3.31, Tomcat 4.1.29. The
system has been solid as a rock the past 3 years until about a month
ago. Since then we have been experiencing intermittent hangs every 2 to
3 days.

 

When the hang occurs, running "apachectl stop" shuts down all the httpd
processs except for the parent process. Attempting a forcefull kill of
the process has no effect - it refuses to go away. Since not all the
processes don't go away, restarting Apache has no effect.

 

Looking at the logs....

 

The Access log

When I look at the access log, there is the obvious gap between the time
when Apache hangs and the system is rebooted.

 

The error log

The last time Apache hung, about 35 minutes after the server quit
responding, I saw the "reached Max Clients" in the error log. Max
Clients was set to 250. So it appears to me that the "Max Client" issue
is a result and not the cause of the hang. Prior errors didn't seem to
indicate anything definitive. The LogLevel was set to "warn", and today
I set it to info too see if I can get Apache to tell me more about
what's going on.

 

Also worth noting is that when the hang occurs, not only are the servlet
pages not available, but also the static "Apache Server" web page is not
available either.

 

Other than Apache being hung, the other services on the system appear to
be working fine- the issue is specific to the Apache/Tomcat server. At
the time of the hang, other protocols such as ICMP, DNS, SMTP, and SSH
all seem to be ok.

 

Has anyone experienced a problem similar to this? Has anyone seen any
good troubleshooting tips for cases such as this? I've made a first pass
through bugzilla without anything definitive showing up.

 

 


Mime
View raw message