About the LVS subproject - TCPHA

[cheaney Chen <cheaney.chen@xxxxxxxxx>]

Hi,

   Has anyone tried to use TCPHA ? I've met some trouble about that.

   My cluster has the following configuration:

I have one front-end server and two back-end servers.

FE: 
   VIP 192.168.20.21   eth0:1
   RIP 192.168.20.3     eth0

BE:
   VIP 192.168.20.21  (tunl0,non-ARP)
   RIP 192.168.20.5    eth0

   VIP 192.168.20.21  (tunl0,non-ARP)
   RIP 192.168.20.7    eth0

And I set the configure file as follows
   tcpha_fe {
      raddr = 192.168.20.3
      vaddr = 192.168.20.21
      port = 80
      redirectaddr = 192.168.20.3
      redirectport = 8080
      server = 192.168.20.5 666
      server = 192.168.20.7 666
      rule = 192.168.20.5 gif $
      rule = 192.168.20.7 .*
}

   tcpha_be{
     raddr = 192.168.20.5
     uaddr = 192.168.20.25
     vaddr = 192.168.20.21
     port = 666
     redirectaddr = 192.168.20.3
     redirectport = 80
}

And I also start Apache on the BE.

But it can't work.

All requests are redirected to 192.168.20.7 .
And some error messages show on the monitor:

TCPHA_BE : no server found
KERNEL : assertionn (sk->forward_alloc ==0) failed at tcp.c(1853)
KERNEL : assertion (atomic_read(&sk->rmem_alloc==0)) failed at af_inet.c(173)
KERNEL : assertion (sk->forward_alloc == 0 ) failed at af_inet.c(176)

Could anyone give me some idea about the question ?

Best regards,
Cheaney Chen