|
On Mon, Feb 21, 2005 at 04:27:18PM +0000, SUKHWINDER PAL wrote:
> Finally we are able to implement two squid servers
> with LVS-DR method .Now we are trying to implement
> ldirectord to detect fail over of squids.If We use two
> realserver as following :-
>
> On director--------
> VIP------10.11.151.24
> DIP-------10.11.150.98
>
> On Realserver1(Squid)-----------
> RIP----------10.11.150.82
> VIP----------10.11.151.24
>
> On realserver2------------
> RIP----------10.11.150.96
> VIP----------10.11.151.24
>
> Now as in LVS-DR the realserver is serving directly to
> client and not through director.Then how can we
> implement ldirectord on director to check health of
> realservers.
> We downloaded the "forward_shared---" patch & applied
> it on director.Then we changed the default gateway of
> realservers as DIP.And then we issued following
> commands:-
> echo 1 > /proc/sys/net/ipv4/conf/all/forward_shared
> echo 1 > /proc/sys/net/ipv4/conf/DIP/forward_shared
> Then we modified /etc/ha.d/ldirectord on the director
> as follows:-
> virtual=10.11.151.24:3128
> real=10.11.150.82:3128 gate
> real=10.11.150.96:3128 gate
> fallback=127.0.0.1:3128
> quiescent=no
> scheduler=wrr
> protocol=tcp
> As on the man page of ldirectord service "squid" was
> not there ,we have not defined service field.
> After this we started all the services like
> ipvsadm,ldirectord,squid on director & realserver
> respectively.But whenever any of the squid server
> goes down its entry is not removed from the ipvsadm
> table.And client connected to the failed server does
> not get connected to the working squid automatically.
> What could be the problem sir.
Have you tried using a connect check on port 3128,
or a http negotiate check on 3128, or perhaps patching
in a more specific squid check and posting the patch here?
--
Horms
|
