https://issues.apache.org/bugzilla/show_bug.cgi?id=55422
Christopher Dillard <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Resolution|INVALID |WONTFIX --- Comment #4 from Christopher Dillard <[email protected]> --- Here's an inelegant workaround that still allows for persistent connections to the back-end: Make multiple named load balancers, each with exactly one BalancerMember, but all of which point to the same origin server, e.g.: <Proxy balancer://origin1> BalancerMember https://192.168.1.200 </Proxy> <Proxy balancer://origin2> BalancerMember https://192.168.1.200 </Proxy> Then make multiple (nearly identical) VirtualHosts with different ServerNames, each using a different balancer: <VirtualHost *:443> ... ServerName dmz.example.com ... ProxyPass /foo balancer://origin1/foo </VirtualHost> <VirtualHost *:443> ... ServerName 192.168.1.100 ... ProxyPass /foo balancer://origin2/foo </VirtualHost> This works around this bug because the way balancers are coded, each gets its own worker connection pool, and the VirtualHosts route to different balancers. So there won't be any connection sharing with requests to a different host name. The down side is that this requires a longer, more error-prone config file. And it requires loading up all the balancer machinery. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
