I need help to decide what is the right method (not iRule in my case)
I have four servers in the pool.
Due to the server's performance, I have a limit of 15 connections per server.
I have a queue time-out 10sec
The first server is the main server. What I need is the following:
The first server will reach the limit of 15 connections
that the traffic will go to the second server, then the third, then fourth
So basically, the load balancing will be
Server 1 > reach 15 connections ? Yes.
Send traffic to Server 2, etc.
Can it possibly be done by the load-balancing method? LTM Policy?
Solved! Go to Solution.
@MaxMedov You should be able to achieve this by setting a connection limit on the pool members and then setting up priority group activation. Once the connection limit is hit that pool member would be considered down which then would cause priority activation to kick in and the next pool member would receive the overflow traffic. The following article goes over priority group activation a bit to assist you in that configuration.
Hi, @Paulius thank you for your quick response.
I know about priority groups, but from what I know and what existed in the documentation, this option will work when the poll member is unavailable, which means he won't reply to the health check or will be disabled. Are you sure it's working the same as the connection limit per pool member?
@MaxMedov I am definitely sure that if you set a connection limit on the pool member it will use priority group activation. So priority group activation will kick in when a status other than green kicks in, in this case the connection limit, and then the next server will take over and not just health check failures.
Thank you @Paulius I test it, and really the status changes to
Unavailable (Enabled) - The pool member's connection limit has been reached
I configured two servers in the pool
server 1 - connection limit 2 | priority group 20
server 2 - connection limit 1 | priority group 10
What expected are two first connections will go to server 1
The following connection to server 2
When one connection in server one is disconnected, the next connection will go to server one again.
In my test, all the connections go to Server 1
*I disabled the persistence of src_ip
@MaxMedov That is correct, I believe because you have your session open and the server is responding it will maintain your connection to it unless it doesn't respond which at that point you will establish a new connection which will balance you to the new pool member that is active because of priority group activation. As an example, if you have the connection limit set to 5 and users 1-5 come in forming a single connection the connection limit will be met and the next set of users will go to the next pool as long as the connection limit remains at the limitation of 5 connections, once the next connection limit is hit the next new connection will move to the next pool because it will now be active and available to receive these connections. In my opinion I wouldn't perform this priority group method and would simply configure least connection (node) as the balancing method and have all servers available in the pool so that you have all your available resources allowed from the beginning instead of the servers sitting and not handling any connections.