11-25-2020 08:27 PM
I'm attempting to setup some DTC records but am stumped how to acheive a desired Healthcheck config.
== The big picture
We have some 'on-prem cloud' services (Kubernetes, et.al) which host some foundational services. We want to use IB DTC to provide the ext. loadbalancer work so clients can hit an appropriate node.
== The problem
Some of the native k8s services are running as a daemonset (on all hosts) which is an easy target to gauge a hosts liveness (keeping a DTC record active for this node) but it's not good enough for service readiness. In a shared service env. like K8s some workloads are only active on a subset of the entire pool members. If these workloads are not HTTP/HTTPS most of the time you have a single port open per pool member which listens for traffic. Since the healthcheck configuration is at the pool level I cannot align the LDBN patterns on routing to specific nodes with the workload running on them.
== Desired solution
Unless I missed something (... entirely possible), I'd really like the ability to define healthchecks at the ldbn / pattern level. This would allow me to use hosts as shared resources with dynamic routing needs.