I have an application which has the specific requirements:
Two nodes in the pool.
Send all requests to only one node during normal operation (hence using priority group activation).
If the favoured node fails, then send all requests to the second node in the pool.
I understand there are implications here around availability, capacity etc, but the application team is aware of this and specifically want this setup. If the favoured node comes back online in the above example, the application team want the F5 to continue sending all requests to the second node to avoid outages.
My issue with the above is this:
If the favoured node comes back online, it will be ignored and never used.
Once this happens, if the second node in the pool fails (after it has already been selected as the sole node to use), traffic will not automatically be redirected back to the favoured node and no node will be selected.
I was thinking there may be some way to achieve the automatic redirection back the favoured node using “Action on service up”, but sadly no event is available. Alternatively, it might be possible by programatically modifying the pool member priority values, but I’m not too sure of how this might be implemented.
Is there any way I can achieve the above functionality automatically on the F5?
If the favoured node comes back online in the above example, the application team want the F5 to continue sending all requests to the second node to avoid outages.
If the favoured node comes back online in the above example, the application team want the F5 to continue sending all requests to the second node to avoid outages.
That’s a really interesting link and something I was unaware of, so thanks for posting! After reading through the entirety of the article, it seems this may not be suitable for situations requiring prioritisation of nodes within the pool though? This is the requirement we are facing, so we do need some way to cater for this. Are there any options with this in mind?
That’s a really interesting link and something I was unaware of, so thanks for posting! After reading through the entirety of the article, it seems this may not be suitable for situations requiring prioritisation of nodes within the pool though? This is the requirement we are facing, so we do need some way to cater for this. Are there any options with this in mind?