For more information regarding the security incident at F5, the actions we are taking to address it, and our ongoing efforts to protect our customers, click here.

Forum Discussion

F5FunTimes_2847's avatar
F5FunTimes_2847
Icon for Nimbostratus rankNimbostratus
Aug 19, 2016

Active Connections Never Bleed Off of Node

I have an F5 BIG-IP sitting in front of two T1 web servers that are configured to round robin traffic. Whenever I want to perform maintenance on one of the two web servers I go into the pool and disable the node which causes new connections to be sent to the other web server and active connections to stay on the node that was disabled.

 

My issue is that users are making calls to our APIs every 5 seconds in some instances and these active connections never bleed off of the node that is disabled (I waited a day once and the same 20 connections were still pinned).

 

Is there a way to safely send the active traffic over to the other web server without causing exceptions to be seen by the end user?

 

thank you

 

application delivery
BIG-IP
devops

2 Replies

  • OTS02's avatar
    OTS02
    Icon for Cirrus rankCirrus
    Aug 19, 2016

    Hi F5FunTimes,

     

    You said you disabled the pool member. Did you really mean that you 'Forced Offline'?

     

  • Vijay_E's avatar
    Vijay_E
    Icon for Cirrus rankCirrus
    Aug 19, 2016

    The connections are probably not closing because of the incessant API calls. I would recommend the following:

     

    1. Forced Offline option.
    2. Use OneConnect with /32 netmask. This will force the F5 to make a load balancing decision per HTTP request instead of the default load balancing decision per TCP connection and hence, move the HTTP API call after the pool member is marked "Forced Offline".