F5 load balancing SMTP to 2 groups of servers

can you try something like this? this is for http.

[root@ve1023:Active] config  b virtual bar list
virtual bar {
   snat automap
   pool foo1
   destination 172.28.19.79:80
   ip protocol 6
   rules myrule
}
[root@ve1023:Active] config  b pool foo1 list
pool foo1 {
   monitor all http
   members 200.200.200.101:80 {}
}
[root@ve1023:Active] config  b pool foo2 list
pool foo2 {
   monitor all http
   members 200.200.200.102:80 {}
}
[root@ve1023:Active] config  b rule myrule list
rule myrule {
   when CLIENT_ACCEPTED {
        if {[active_members [LB::server pool]] < 1} {
                pool foo2
        }
}
}

Thanks. I will give it a try!

Hi Larry G,

 

 

If you go with the scenario that you have given you could suffer some adverse affects (depending on your traffic load and your server abilities).

 

 

If you have four servers in each group.

 

1 fails - traffic is redistributed to the remaining 3.

 

2 fail - traffic is redistributed to the remaining 2.

 

3 fail - traffic traffic goes the the remaining server and possibly overload the server and/or cause the responsiveness to degrade (again, depending on your traffic load).

 

 

I would suggest putting all eight servers in a single pool and using Priority Activation Groups with a set of requirements that can be easily adjusted to fit your needs.

 

 

Example:

 

Server 1 - Group 2

 

Server 2 - Group 2

 

Server 3 - Group 2

 

Server 4 - Group 2

 

Server 5 - Group 1

 

Server 6 - Group 1

 

Server 7 - Group 1

 

Server 8 - Group 1

 

 

NOTE: Group Number of the highest number is active, so set your "Primary" Servers in the highest group number.

 

 

Set Priority Activation Groups to less than 4. This will keep 4 servers active at all times.

 

 

If a server in Group 2 (Active Server Set) fails, a single server from Group 1 would be made active to keep the load distribution equal (no traffic spikes or performance degradation).

 

If two servers in Group 2 fail, two servers are activated in Group 1 and so on. When the servers in Group 2 are recovered, the traffic going to the server(s) in Group 1 will "finish up" the connections that they have and all new traffic would go back to normal. All automagically.

 

 

Just a idea to try. Hope this helps.

hi Michael, good point. 🙂

by the way, just curious when a number of active pool member in priority group 2 is less than 4, will priority group 1 bring only some of its pool member (not all the member)?

this is my test. could you please advise if i missed anything?

[root@ve1023:Active] config  b virtual bar list
virtual bar {
   snat automap
   pool foo
   destination 172.28.19.79:80
   ip protocol 6
   rules myrule
   profiles {
      http {}
      tcp {}
   }
}
[root@ve1023:Active] config  b pool foo list
pool foo {
   min active members 2
   monitor all http
   members {
      200.200.200.101:80 {
         priority 2
      }
      200.200.200.102:80 {
         priority 2
      }
      200.200.200.111:80 {
         priority 1
      }
      200.200.200.112:80 {
         priority 1
      }
   }
}
[root@ve1023:Active] config  b pool foo|grep -i pool\ member
+-> POOL MEMBER foo/200.200.200.101:80   active,up
+-> POOL MEMBER foo/200.200.200.102:80   active,up
+-> POOL MEMBER foo/200.200.200.111:80   active,up
+-> POOL MEMBER foo/200.200.200.112:80   active,up

[root@ve1023:Active] config  ab -c1 -n100 http://172.28.19.79/
...

[root@ve1023:Active] config  b pool foo
POOL foo  LB METHOD round robin   MIN/CUR ACTIVE MEMBERS 2/4
|     (cur, max, limit, tot) = (0, 2, 0, 100)
|     (pkts,bits) in = (500, 287200), out = (500, 597200)
+-> POOL MEMBER foo/200.200.200.101:80   active,up
|   |     session enabled    priority 2    ratio 1
|   |     (cur, max, limit, tot) = (0, 1, 0, 50)
|   |     (pkts,bits) in = (250, 143600), out = (250, 366000)
|   |     requests (total) = 50
+-> POOL MEMBER foo/200.200.200.102:80   active,up
|   |     session enabled    priority 2    ratio 1
|   |     (cur, max, limit, tot) = (0, 1, 0, 50)
|   |     (pkts,bits) in = (250, 143600), out = (250, 231200)
|   |     requests (total) = 50
+-> POOL MEMBER foo/200.200.200.111:80   active,up
|   |     session enabled    priority 1    ratio 1
|   |     (cur, max, limit, tot) = (0, 0, 0, 0)
|   |     (pkts,bits) in = (0, 0), out = (0, 0)
|   |     requests (total) = 0
+-> POOL MEMBER foo/200.200.200.112:80   active,up
    |     session enabled    priority 1    ratio 1
    |     (cur, max, limit, tot) = (0, 0, 0, 0)
    |     (pkts,bits) in = (0, 0), out = (0, 0)
    |     requests (total) = 0

[root@ve1023:Active] config  b pool foo|grep -i pool\ member
+-> POOL MEMBER foo/200.200.200.101:80   inactive,down
+-> POOL MEMBER foo/200.200.200.102:80   active,up
+-> POOL MEMBER foo/200.200.200.111:80   active,up
+-> POOL MEMBER foo/200.200.200.112:80   active,up

[root@ve1023:Active] config  ab -c1 -n100 http://172.28.19.79/
...

[root@ve1023:Active] config  b pool foo
POOL foo  LB METHOD round robin   MIN/CUR ACTIVE MEMBERS 2/3
|     (cur, max, limit, tot) = (0, 3, 0, 100)
|     (pkts,bits) in = (500, 287200), out = (500, 551368)
+-> POOL MEMBER foo/200.200.200.101:80   inactive,down
|   |     session enabled    priority 2    ratio 1
|   |     (cur, max, limit, tot) = (0, 0, 0, 0)
|   |     (pkts,bits) in = (0, 0), out = (0, 0)
|   |     requests (total) = 0
+-> POOL MEMBER foo/200.200.200.102:80   active,up
|   |     session enabled    priority 2    ratio 1
|   |     (cur, max, limit, tot) = (0, 1, 0, 33)
|   |     (pkts,bits) in = (165, 94776), out = (165, 152592)
|   |     requests (total) = 33
+-> POOL MEMBER foo/200.200.200.111:80   active,up
|   |     session enabled    priority 1    ratio 1
|   |     (cur, max, limit, tot) = (0, 1, 0, 34)
|   |     (pkts,bits) in = (170, 97648), out = (170, 157216)
|   |     requests (total) = 34
+-> POOL MEMBER foo/200.200.200.112:80   active,up
    |     session enabled    priority 1    ratio 1
    |     (cur, max, limit, tot) = (0, 1, 0, 33)
    |     (pkts,bits) in = (165, 94776), out = (165, 241560)
    |     requests (total) = 33

thanks!