Forum Discussion

SKS_198176's avatar
SKS_198176
Icon for Nimbostratus rankNimbostratus
Aug 31, 2022
Solved

Priority group and least connection.

I have recently started my F5 journey and have a question.

Let's suppose that I have a pool with 2 members that is configured as following:
Load Balancing Method: least connection
Priority Group Activation: less than 1
Member A priority 2 (connection limit 80)
Member B priority 1 (connection limit 80)
How is the load balancing going to work in this case? For example, will it be load balanced based on the least connection or the priority group?
If the decision is based on the priority group, does the F5 route the traffic to the server in the lower priority group when the connection limit is reached?

  • CA_Valli's avatar
    CA_Valli
    Sep 01, 2022

    Hi, source address persistency is not a good idea if all traffic comes from a single IP address. 

    What is your expected scenario? Why do you think you will need persistency?

    If there is a persistency record in the table, F5 basically bypasses load balancing decision and forwards the packet to the member bound to the record. This does not really change much in your scenario since PG2 already only contains one member. 

    When Member A has reached full capacity and a new client connection is received before persistency record expires, I believe LTM will try to forward this connection to Member A and you will hit a LB_FAILED condition.

    You can see in vs and pool statistics if there is any failed LB or resets, or script an iRule that logs a message when this condition is met.

4 Replies

  • With this setup, LTM will normally select Member A for traffic forwarding 100% of the time. When Member A reaches connection threshold, member B will be included in forwarding decisions and due to leastconn method it will receive 100% of connection that exceed Member A capacity. 

  • Hi CA_Valli,

    Thanks for the reply, further refinement to the above configuration

    What will happen if I configure a source_addr persistence with 120 sec timeout setings ( which means that Member A will receive all the connection for 2 minutes) as it has the highest priority settings.

    with the least connection LB method configured with 80 max connection setting, will the LTM reset the 81st connection received within 120 sec time period.

    • CA_Valli's avatar
      CA_Valli
      Icon for MVP rankMVP

      Hi, source address persistency is not a good idea if all traffic comes from a single IP address. 

      What is your expected scenario? Why do you think you will need persistency?

      If there is a persistency record in the table, F5 basically bypasses load balancing decision and forwards the packet to the member bound to the record. This does not really change much in your scenario since PG2 already only contains one member. 

      When Member A has reached full capacity and a new client connection is received before persistency record expires, I believe LTM will try to forward this connection to Member A and you will hit a LB_FAILED condition.

      You can see in vs and pool statistics if there is any failed LB or resets, or script an iRule that logs a message when this condition is met.