Forum Discussion

dcarterjr's avatar
dcarterjr
Icon for Altostratus rankAltostratus
Feb 01, 2023
Solved

Add LTM to existing HA pair

I need to add 2 LTMs to an exsiting HA pair, and create a second traffic group in that device group.  The second pair are in a different location.  I've looked at many F5 articles, and it looks doable, but I haven't found one that talks specifically about adding these extra devices to the HA pair.  I'd appreciate any input.  These are FIPS boxes, if that makes any difference.  7200s, Version 14.5.1.2.  Thanks

  • dcarterjr I believe the best route would be to create the new traffic group on the existing pair first to confirm it functions. After that you should be able to add in the other devices to the HA pair the same as any other F5 HA setup. This article might help you with your configuration.

    https://clouddocs.f5.com/training/community/adc/html/class1/module1/lab4.html

    The only concern I would have is latency between the 4 devices because those other devices are at a different location. As long as you can extend your subnets for the existing pair to the other location for the two new devices it should be possible to configure this setup.

10 Replies

  • dcarterjr I believe the best route would be to create the new traffic group on the existing pair first to confirm it functions. After that you should be able to add in the other devices to the HA pair the same as any other F5 HA setup. This article might help you with your configuration.

    https://clouddocs.f5.com/training/community/adc/html/class1/module1/lab4.html

    The only concern I would have is latency between the 4 devices because those other devices are at a different location. As long as you can extend your subnets for the existing pair to the other location for the two new devices it should be possible to configure this setup.

    • dcarterjr's avatar
      dcarterjr
      Icon for Altostratus rankAltostratus

      Paulius.

      Thanks for the info.  I worry a little about latency also, but the second pair and the new traffic-group are so we cam move applications to the new location along with moving them to the new traffic group to prevent application latency.  The first pair does not need to failover to the second group, just sync.  I'll take a look at the aricle and let you know how it goes.  

      Dave

      • Paulius's avatar
        Paulius
        Icon for MVP rankMVP

        dcarterjr If the new pair is not a failover pair for the first pair I would configure those new ones as their own HA pair and then copy the configuration over manully by using CLI list commands or creating an SCF file and grabbing the appropriate configuration and moving it over to the new units. You can do this by copying the configuration in question and then typing the following command which will allow you to paste in the configuration and save it to the new pair. Keep in mind that if any of the configuration is hashed you will need to configure the exact same master key on the new pair for the configuration to load properly.

        load sys config from-terminal merge

        It is important that you use this exact command because if not you could overwrite the existing configuration of the new F5 pair. The reason I recommend the process above is because it seems like you are trying to save the leg work of configuring the new HA pair by using the SYNC but you shouldn't do that because these new devices will not be failover devices for the existing HA pair.

  • Great conversation.
    Thanks for so much context dcarterjr and for providing so much help and clarification Paulius.
    Were you ever able to get this working?
    If so...were one or more of the replies here complete enough to *Accept As Solution* ?

    Would love to have this highlighted as a solution for the community if possible.
    Cheers,
    Lief

    • dcarterjr's avatar
      dcarterjr
      Icon for Altostratus rankAltostratus

      The solution provided by Paulius worked well.  Latency was an issue, so we extened the timeout for communication failure between devices from 3 seconds to 10 seconds, which worked well.  It was a good learning experience.  Everyone's help was greatly appreciated.