Forum Discussion

cody8411_186897's avatar
cody8411_186897
Icon for Nimbostratus rankNimbostratus
Jul 28, 2016

Active/Standby load balancing dynamically with LTM

I'm not sure if I'm using the right terminology, but some application folks want to set up identical web servers where they can test upgrades and changes to server A while maintaining production traffic to server B.

LTM is currently directing traffic for the application using an iRule and matching URIs so it looks like:

www.company.com/app1 -> App1_Pool -> Node1 & Node2

Is there something I can do to intelligently determine which node (or a different pool) this lands on without administrator intervention? Ideally something that they can trigger themselves when they are ready to upgrade their apps.

www.company.com/app1 -> App1_Pool -> Node 1 (if active) or Node 2 (if active)

I thought of them possibly disabling the site in IIS which would make the port monitor go down and stop traffic flow but they want to still access the app in a testing capacity so that likely won't work.

Any help is appreciated, thanks!

  • Priority groups. You can assign nodes in a pool to different priority groups so that some nodes will only be used if others with a better priority are down.

     

  • Priority groups. You can assign nodes in a pool to different priority groups so that some nodes will only be used if others with a better priority are down.

     

    • cody8411_186897's avatar
      cody8411_186897
      Icon for Nimbostratus rankNimbostratus

      Is there some other way I can monitor those devices to make it so they can trigger the priority in those pools? For instance, if I just use a basic TCP port monitor, those ports will never go down or be disabled.

       

    • ekaleido_26616's avatar
      ekaleido_26616
      Icon for Cirrocumulus rankCirrocumulus

      You can create a custom health monitor such that they can modify something to cause the server to be removed from the pool, if that's what you're after.

       

    • cody8411_186897's avatar
      cody8411_186897
      Icon for Nimbostratus rankNimbostratus

      Great idea, I'll start to look at making a GET to return a string and take the the pool member down if it doesn't exist. That way they can remove the string on the passive node.

       

  • Priority groups. You can assign nodes in a pool to different priority groups so that some nodes will only be used if others with a better priority are down.

     

    • cody8411_186897's avatar
      cody8411_186897
      Icon for Nimbostratus rankNimbostratus

      Is there some other way I can monitor those devices to make it so they can trigger the priority in those pools? For instance, if I just use a basic TCP port monitor, those ports will never go down or be disabled.

       

    • ekaleido's avatar
      ekaleido
      Icon for Cirrus rankCirrus

      You can create a custom health monitor such that they can modify something to cause the server to be removed from the pool, if that's what you're after.

       

    • cody8411_186897's avatar
      cody8411_186897
      Icon for Nimbostratus rankNimbostratus

      Great idea, I'll start to look at making a GET to return a string and take the the pool member down if it doesn't exist. That way they can remove the string on the passive node.