Forum Discussion

forsan_102218's avatar
forsan_102218
Icon for Nimbostratus rankNimbostratus
May 21, 2015

Decision making in a Active Active senario

Hi,

 

I have an issue I need to find a solution for. In a setup I have one redundant link between the Active/Standby pair. Both management and our HA VLAN goes over this link. The F5:s are in different datacenter. This link is on one ISP but on two different pats.

 

In the worst case senario and this LACP link goes down the BigIPs both go Active. When the link returns to normal operations I want the BigIP on site1 to be active and the one on site2 to be standby. But I get it the other way around. What I have found is that this probably is caused due to site2 BigIP has a higher IP address on the management interface compared to the one on site1. The management addresses are on different subnets so its not just to change the IP address.

 

Does anyone has solution for my problem? I don't want to use the Auto Fallback function because I want to do a controlled manual failover if the primary one fail.

 

Best Regards Andréas

 

5 Replies

  • Use HA Groups, a much better way to manage failover triggering and state

     

  • When the link returns to normal operations I want the BigIP on site1 to be active and the one on site2 to be standby. But I get it the other way around.

     

    what version are you using? starting in 11.4.0, ha order failover can be configured. i never tested (active/active to active/standby) but thinking it may be usable.

     

    Ordered failover lists for Sync-Failover device groups

     

    An alternative to the new load-aware failover feature is an ordered list. For each traffic group in the device group, you can specify a list of devices that you want to be the next-active device for the traffic group if failover occurs. If the first device in the list is unavailable, the BIG-IP system chooses the next device in the list to be the next-active device, and so on. This feature is more useful for homogeneous, rather than heterogeneous, platforms in device groups.

     

    Release Note: BIG-IP LTM and TMOS 11.4.0

     

    https://support.f5.com/kb/en-us/products/big-ip_ltm/releasenotes/product/relnote-ltm-11-4-0.html

     

    • forsan's avatar
      forsan
      Icon for Altostratus rankAltostratus
      Hi, Thank you for your answer. I'm running 11.5.2 and I have configured the HA Order in the traffic-group. I have both units in the selected list with the loadbalancer in site1 at the top and the site2 as second in the list. It looks in the log that its chooses the site2 due to "best load". I don't have logs from the device in front of me right now so I can't check right now. I can check on Wednesday an get back with the log message. Regards Andréas
  • When the link returns to normal operations I want the BigIP on site1 to be active and the one on site2 to be standby. But I get it the other way around.

     

    what version are you using? starting in 11.4.0, ha order failover can be configured. i never tested (active/active to active/standby) but thinking it may be usable.

     

    Ordered failover lists for Sync-Failover device groups

     

    An alternative to the new load-aware failover feature is an ordered list. For each traffic group in the device group, you can specify a list of devices that you want to be the next-active device for the traffic group if failover occurs. If the first device in the list is unavailable, the BIG-IP system chooses the next device in the list to be the next-active device, and so on. This feature is more useful for homogeneous, rather than heterogeneous, platforms in device groups.

     

    Release Note: BIG-IP LTM and TMOS 11.4.0

     

    https://support.f5.com/kb/en-us/products/big-ip_ltm/releasenotes/product/relnote-ltm-11-4-0.html

     

    • forsan's avatar
      forsan
      Icon for Altostratus rankAltostratus
      Hi, Thank you for your answer. I'm running 11.5.2 and I have configured the HA Order in the traffic-group. I have both units in the selected list with the loadbalancer in site1 at the top and the site2 as second in the list. It looks in the log that its chooses the site2 due to "best load". I don't have logs from the device in front of me right now so I can't check right now. I can check on Wednesday an get back with the log message. Regards Andréas