On 05/05/13 11:19, Hugo Silva wrote:
> To begin with, I forgot to mention that we're on 1.5.d18.
> 
> We use inspect-delay for something else (that works), as shown below:
> 
> 
> backend xmpp-tcp
>     mode                    tcp
>     source                  x.x.x.x
> 
>     option                  tcplog
> 
>     timeout                 tunnel 70000
> 
>     server                  livefire     y.y.y.y:5222 check inter 10s
>     server                  stagingfire  z.z.z.z:5222 check inter 10s
> 
>     acl                     valid_xmpp_client   payload(0,5) <?xml
>     acl                     xmpp_to_live        payload(0,500) -i to=g
>     acl                     xmpp_to_staging     payload(0,500) -i to=s
> 
>     tcp-request             inspect-delay 7s
>     tcp-request             content reject unless valid_xmpp_client
> 
>     use-server              livefire       if xmpp_to_live
>     use-server              stagingfire    if xmpp_to_staging
> 
> 
> We've tried variations of -m sub, -m reg, and commenting out the
> valid_xmpp_client check in order to just send LIVE or STAGING (adapting
> the xmpp_to* acls) and match that exactly but it didn't work either, and
> that's when we thought this is the wrong way to go about it.


Any ideas on this? We need to revert to something usable asap, if we
can't make this work then we'll just assign a different TCP port to the
2nd xmpp server. But that doesn't seem like the most correct option.

Hugo


Reply via email to