Lighttpd

The below config works fine with 1.4.8, traffic is distributed evenly among the FCGI backends. I'm handling around 200 requests per second on this setup. When a backend disconnects, the traffic is moved to other backends cleanly.

After upgrading to 1.4.9 the traffic was distributed only (or mostly) to the backends on 192.168.209.190 and no traffic went to the 127.0.0.1 backends. Furthermore when one of the backends disconnected the customers got a Error 500 page instead of being redirected to other backends. And in the logs I saw a lot of this:
2006-01-19 15:47:22: (mod_fastcgi.c.2672) fcgi-server re-enabled: 192.168.209.190 9910
2006-01-19 15:47:24: (mod_fastcgi.c.2775) establishing connection failed: Connection timed out port: 9910
2006-01-19 15:47:26: (mod_fastcgi.c.2775) establishing connection failed: Connection timed out port: 9910

So basically the upgrade caused only failures and I downgraded for now. Please advise.

server.modules = ( "mod_rewrite", "mod_fastcgi", "mod_accesslog" )
server.document-root = "/opt/lighttpd/wwwroot" 
server.indexfiles = ( "index.html", "index.htm" )
server.port = 80
server.bind = "216.139.209.104" 
server.errorlog = "/opt/lighttpd/error.log" 
server.event-handler = "linux-sysepoll" 
server.max-worker = 8
debug.log-file-not-found = "enable" 
#accesslog.filename = "/opt/lighttpd/access.log" 

fastcgi.server = (
  "/main.fcgi" => (
    "w8one" => (
      "host" => "192.168.209.190",
      "port" => 9910,
        "check-local" => "disable",
        "disable-time" => 5
    ),
    "w8two" => (
      "host" => "192.168.209.190",
      "port" => 9911,
        "check-local" => "disable",
        "disable-time" => 5
    ),
    "main" => (
      "host" => "127.0.0.1",
      "port" => 9910,
        "check-local" => "disable",
        "disable-time" => 5
    ),
    "duplicate" => (
      "host" => "127.0.0.1",
      "port" => 9911,
        "check-local" => "disable",
        "disable-time" => 5
    )
  )
)

url.rewrite = (
  "^(/.*)$" =>  "/main.fcgi$1",
)

-- wsobczuk