Forum Discussion

lfhx_54443's avatar
lfhx_54443
Icon for Nimbostratus rankNimbostratus
Aug 19, 2009

Question about persistence and irule

Hi friends,

 

I have used LTM some years but I happened to meet some questions which need your help.

 

 

Q1: what's the priority between iRule and Persistence?

 

Say I have persistence p1 and irules r1 on the same VIP

 

This VIP has 2 members, m1 and m2

 

p1 enable Source Address Affinity, thus if I use the same client to connect, all the request will got to m1

 

while r1's logic is forward all the request to m2

 

Thus, who would affect finally, p1 or r1?

 

 

Q2: For persistence

 

The same background in Q1,

 

p1 enable Source Address Affinity, thus if I use the same client to connect, all the request will got to m1

 

But after I disable m1 in LTM, but m1 works there, the request would still go to m1, is that correct?

 

And what about if I shutdown m1, will the request always go to m1 or after it found it can't connect, it would come to m2?

 

Thanks!

2 Replies

  • The iRule takes precedence over the virtual server and/or pool configuration. If the iRule doesn't change the pool or persistence method, then the VIP's persistence setting would still be used.

     

     

    If a request is made with a persistence record pointing to a pool member that has been marked down by a monitor the request will be re-load balanced according to the load balancing algorithm of the pool. If the node has been disabled, but the pool member still responds to monitor requests requests with a valid persistence record pointing to that pool member will continue to be sent to the pool member.

     

     

    If you set the node to forced offline, only active TCP connections will continue to be allowed.

     

     

    I don't think a persistence record will be invalidated if the pool member responds to a monitor within the timeout but doesn't respond to a client request.

     

     

    Aaron