Hi all,
when using "mod_proxy_ajp" of Apache2 2.2.16 in combination with two Apache
Tomcat 6.0.29 servlet containers, we often see the following entries within the
Apache2 error log of the virtual host:
[Thu Nov 25 16:17:26 2010] [error] ajp_read_header: ajp_ilink_receive failed
[Thu Nov 25 16:17:26 2010] [error] (120006)APR does not understand this error code: proxy: read response failed from 10.35.32.241:8009 (rb-wcmstc2.de.bosch.com)
When looking in the central Apache2 error log, we see:
[Thu Nov 25 16:17:26 2010] [error] (70007)The timeout specified has expired: ajp_ilink_receive() can't receive header
Additionally, we sporadically see connection switches from one Tomcat
to the other without any understandable reasson (here from "rb-wcmstc2" to "rb-wcmstc1"):
10.35.32.123 - - [25/Nov/2010:16:17:31 +0100] "POST /servlet/ClientIO/5cmxicinliar/s6/234 HTTP/1.1" 500 1255 "-" "Jakarta Commons-HttpClient/3.1" "
JSESSIONID=68BE78C5C6A2B9756CED86543095C221.rb-wcmstc2" "68BE78C5C6A2B9756CED86543095C221.rb-wcmstc2" "JSESSIONID=C9F5C78771552AF9C15F1D0064099948.
rb-wcmstc1; Path=/" 4784
Might the below timeout be set to short (in combination with the aboe "timeout" error log entry)?
The current "mod_proxy_ajp" configuration looks like this:
<Proxy balancer://fs4server>
# One tomcat node has been outcommented due to the following ticket: Help-20100510-651
BalancerMember ajp://rb-wcmstc1.xxx.dot.com:8009 loadfactor=100 retry=10 route=rb-wcmstc1
BalancerMember ajp://rb-wcmstc2.xxx.dot.com:8009 loadfactor=100 retry=10 route=rb-wcmstc2
ProxySet stickysession=JSESSIONID|jsessionid
ProxySet lbmethod=byrequests
ProxySet scolonpathdelim=On
#ProxySet nofailover=On
ProxySet timeout=45
</Proxy>
Should we switch to "mod_proxy_http"?