when_lb_failed

Can

 

 

Would a generic pool setting of Action on Service Down set to Reselect help here or do you specifically need to target only those hosts from 192.168.1.0 and 192.168.2.0?

 

 

Hope this helps,

 

N

Hi Nathan,

 

 

Actually I need to indicate each remote network subnet and I need to load balance according to the networks. while clients are coming run two tcp port like 41000 and 56000 and these ports need to run on the same node.Therefore, I have configured the source adress affinity as match across services.and now as mentioned above how can I load balance via an irule. as you understand each node has both the 41000 and the 56000 ports.That's why there are two virtual server which have the same ip but just ports different like 41000 and 56000 so there are two pool and each pool member include 192.168.1.10:41000 and 192.168.1.11:41000 and other pool include 192.168.1.10:56000 and 192.168.1.11:56000

 

 

if clients remote address match the 192.168.1.0 or 192.168.2.0 /24 network redirect to node1 else redirect to node2 but if node1 is down all active and new connections redirect to nod2 or if node2 is down all active and new connections redirect to node1

 

 

I guess We need to use when client accepted and when lb failed parameters.

 

 

Thank you in advance.

Will this work for you then?

 

 

Create a Data Group for 192.168.1.x and 192.168.2.x clients:

 

 

class 192_group {

 

{

 

network 192.168.1.0/24

 

network 192.168.2.0/24

 

}

 

}

 

 

Create an iRule:

 

 

when CLIENT_ACCEPTED {

 

 

Check if Client Address is from the specific data group

 

if { [class match [IP::remote_addr] equals $::192_group]} {

 

 

node 192.168.1.10

 

} else {

 

node 192.168.1.11

 

}

 

}

 

 

when LB_FAILED {

 

pool test_pool

 

LB::reselect

 

log local0. "Selected server [LB::server] did not respond. Re-selecting node from test_pool"

 

}

 

 

Most of this taken from: https://devcentral.f5.com/wiki/iRules.node.ashx

 

 

If you're on v9.x then use matchclass rather than class match command.

 

 

Rgds

 

N

what if virtual sever is listening on any port and pool member is also on any port. on pool, assign 2 health monitors which one checkes on port 41000 and the other one is on port 56000.

 

 

for client 192.168.1.0/24 and 192.168.2.0/24, just check IP::client_addr and send to specific pool member if it is up.

Hi Nitass,

 

 

let me be clear actually there are two virtual server which are listening different ports but their ip address same but their pools different because as you mentioned above each node is listening different ports and while a client is requesting to server which is running two different tcp ports and these ports need to be on the same server.therefore there are two different pool.

 

 

regards,

i mean something like this. is it what you are asking?

[root@ve10:Active] config  b virtual bar list
virtual bar {
   snat automap
   pool foo
   destination 172.16.1.1:any
   ip protocol 6
   rules myrule
}
[root@ve10:Active] config  b pool foo list
pool foo {
   monitor all tcp41000 and tcp56000
   members {
      192.168.1.10:any {}
      192.168.1.11:any {}
   }
}
[root@ve10:Active] config  b monitor tcp41000 list
monitor tcp41000 {
   defaults from tcp
   dest *:41000
}
[root@ve10:Active] config  b monitor tcp56000 list
monitor tcp56000 {
   defaults from tcp
   dest *:56000
}
[root@ve10:Active] config  b rule myrule list
rule myrule {
   when CLIENT_ACCEPTED {
   if {[class match -- [IP::client_addr] equals network_class]} {
      if {[LB::status pool foo member 192.168.1.10 0] equals "up"} {
         pool foo member 192.168.1.10 0
      }
   }
}
}
[root@ve10:Active] config  b class network_class list
class network_class {
   {
      network 192.168.1.0/24
      network 192.168.2.0/24
   }
}

Hello Nitass ,

 

 

thank you for your reply. However the services ports are wanted to separate so we cannot set any port on the members.

 

 

regards,