Is it possible to keep Kong DB-Less running, updating the local yaml file in the file system. Then somehow reload that yaml of config while Kong is taking active traffic without interrupting anything? Or does Kong DBLESS when doing a config change(new routes/service/consumers/plugins) require like a rolling redeploy alongside pulling in the new config to not impact hot traffic?
Would make my life easier if Kong could just keep active running for traffic and when a spec gets uploaded and I can refresh the resources based on the new doc impact free somehow(basically Kong just continues doing its thing and uses old YAML resources until new YAML is fully parsed and the resources are ready for use).
Any masterminds behind Kong DB-less have ideas here?
This is functinoality I would like to have
1 Like
This is possible and supported by Kong in DB-less mode, or at lest in a variation without restarting running kong instance.
@jeremyjpj0916 - refer tot his documentation: DB-less and Declarative Configuration - v2.0.x | Kong - Open-Source API Management and Microservice Management
I read that documentation, has no pointers on handling restarts or config changes in flight for hot path traffic and the behavior one should expect from doing so.
And what behavior will that yield on a Kong node handling existing traffic in real time during the moment of config reload time? I have not tested it yet as I am just now beginning my initial analysis into the feasibility of db-less and best practice. Hence this topic. Have not tested and confirmed yet so was looking if anyone has been using db-less and takes hot path traffic and does reloads and has confirmed if traffic from clients gets messed up during that instant in time of the reload.
