Hey everyone,
I would like to know which fail over time is consider “normal”.
I test stolon in Kubernetes and it works really well.
If I do a fail over test (deleting a pod / node) I can see approx. 20 - 40 seconds downtime for an application.
Does anybody got a suggest how I could tweak that or have no failed requests at all?
Best Regards
Patrick
@paprickar 20-40 seconds aren’t too bad for a single primary database like postgres, reducing them a lot will lead to unwanted failovers. To reduce failover times other kind of databases that provide native high availability should be better.
This must be handled on the client side (your application), not on the server side. Like any other client/server interaction you should handle errors and choose what to do (return the error, retry etc…)
Hey @sgotti
thanks for your response I understand.
I also found the settings I was searching for “failInterval” but I will set it with caution.