We’ve configured kong using kubernetes annotated ingress and service resources, so I’m clearly missing something here… despite that config, we don’t truly have high availability, as our APIs become completely inaccessible for a minute or two if the postgres pod is spinning back up in its new home.
And doing node rolls on a minimum monthly basis means we’re guaranteed downtime, which is not good.
Can anyone help me figure out what we’re doing wrong? Can we use ingress/service resources in dbless mode? Or is there a way to make the postgres pod highly available? We’re using the kong helm chart found at https://charts.konghq.com - is there something simple we can do to address this? We don’t have a ton of resources to take on big infra projects, so looking for the easiest way to get to true HA.