https://issues.apache.org/bugzilla/show_bug.cgi?id=48388
--- Comment #8 from metatec...@gmail.com --- Hi, We started encountering the same error after migrating from Solaris 10 to Solaris 11.1, when the Apache server tries to contact a backend on another machine. [Mon Jan 05 10:54:31.905841 2015] [proxy:error] [pid 10503:tid 4] (146)Connection refused: AH00957: HTTP: attempt to connect to 1.1.1.1:8000 (mymachine.mycompany.com) failed [Mon Jan 05 10:54:31.905889 2015] [proxy:error] [pid 10503:tid 4] AH00959: ap_proxy_connect_backend disabling worker for (mymachine.mycompany.com) for 60s [Mon Jan 05 10:54:31.905909 2015] [proxy_http:error] [pid 10503:tid 4] [client 1.1.1.2:43413] AH01114: HTTP: failed to make connection to backend: mymachine.mycompany.com This happens when there is a sudden burst of HTTP connection, typically in peak hours. The burst can be very short, for instance 10-20 requests during a few seconds. In a TCP dump, we see that the local (outgoing) TCP port for the connections gets reused too quickly, sometimes after a few seconds. When the Apache server and the back-end are on the same machine, this error never occurs, probably because a TCP port used by a connection in "TIME_WAIT" is never assigned by the TCP/IP stack as local port. This is most probably a Solaris bug, but for those searching Google for this error will hit this bug description. A potential workaround is to use mod_cluster. Regards. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org For additional commands, e-mail: bugs-h...@httpd.apache.org
