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

DevBabu_124276's avatar
DevBabu_124276
Icon for Nimbostratus rankNimbostratus
Oct 02, 2015

Configuration advice

I have four app servers behind BIG-IP.

 

The app owners restart their services whenever the memory on the box rises to a certain threshold value. Because of this BIG-IP marks the server that is restarted down monitored by TCP.

 

The users connected to the server through load balancer gets some RPC failed error.

 

VIP has source address persistence enabled. Action on service down is none.

 

1) When the server is marked down, does persistence entry gets deleted or remains until idle time expires. 2) When the server is marked down, does the connection entry gets deleted or remains until idle time expires.

 

App owners has said that the user/session information is known to all servers.

 

Is there any way I could send those users to any available servers.

 

1 Reply

  • Hello,

    I'll try to explain and hopefully answer your questions.

    1) When the server is marked down, does persistence entry gets deleted or remains until idle time expires.

    It will be deleted. There aren't any quick solutions to modify this, however, still doable with the use of iRules if really needed. I'm assuming that you like the default behaviour in that regard.

    2) When the server is marked down, does the connection entry gets deleted or remains until idle time expires.

     

    By default, both the clientside and serverside connection records will remain. These records will be retained until connection is resumed, until timed out, or until closed either by server or client.

    What might interest you is a configuration entity "Action on Service Down" which can be found in Pool configuration settings. In your case, you should go to Pool configuration, and select

    Action on Service Down -> Reselect
    . What this will do is let the active connection flow continue by remapping the serverside connection stream to another available pool member. Please note that this solution would not make anything better for services that require persistent/sticky sessions.