TODO

2022.11

--todo:


not applicable / unrelated to istio..
- java -> kotlin & eliminate lombok
- mvn -> gradle
  iirc there's an issue with the migration to gradle due to a dependency on a maven plugin for the front-end stuff (wro4j).

  https://github.com/spring-petclinic/spring-petclinic-cloud appears to be out of sync with,
  or to have diverged from the https://github.com/spring-projects/spring-petclinic:
  - the .less files are now .scss files, and the wro plugin is no longer used (since we're using sass)
  - the java class names are different
  (i suspect these changes were made to support gradle)

--done:
x review the docs/readme learn the basics of using the repo
x setup a local k3d environment with registry, install istio
x review the code
x understand the app and how it functions
x reconfigure gateway routing of service api endpoints in istio terms using a VirtualService
x add prometheus annotations to app deployment manifests so that sidecar can pick up app container metrics
x eliminate unneeded spring cloud dependencies
x learn to build the code, create the images, and push them to a registry that k8s can fetch images from
x deploy the app to k8s (manifests)
x configure/deploy the mysql databases for each service
x update to latest stable version of jdk/jvm - from 11 to 19
x remove spring cloud kubernetes, can revisit that at a later time
x update version of spring/springboot to latest stable: 2.6.3 -> 2.7.5
x read up on spring cloud kubernetes - https://docs.spring.io/spring-cloud-kubernetes/docs/current/reference/html/
x is the manifest role.yaml (that configures a role named "namespacereader" with rolebinding) even needed?
  it is necessary to allow reading the application.yaml info from the configmap.
  see the spring cloud kubernetes project documentation for more information.
x give each service its own service account
x Deploy the app and see it work
x get routing to function
x Get to the bottom of why services often fail to come up or take an inordinate amount of time.
  need to pay closer attention to service resource requests/limits
x Run prometheus, grafana, jaeger & kiali; send traffic to the app; watch kiali dashboard
x Verify/fix trace headers propagation
  x see https://istio.io/latest/docs/tasks/observability/distributed-tracing/overview/
  x replace deprecated spring cloud sleuth with new way to propagate distributed traces
x Verify that app metrics are collected in addition to envoy metrics
x accessing http://localhost/#!/owners/details/6 from a browser
   results in being able to see the full distributed trace,
   but when i use siege i cannot see the backend services in kiali
   - that's because you are calling the frontend url which assumes a browser
     that executes javascript to then call the backend api endpoint.
     what you want is to capture the url of the backend api endpoint
x improve the build files - i see a lot of repetition; perhaps even incorrect parent-child relationship configuration
x Remind yourself how to work with maven multi-module projects.
  I suspect that many of the common dependencies in the child modules can be pulled into the parent.
x Replace build system: mvn -> gradle
  - there may be an issue with the wro plugin for gradle and the converstion of .less files to .css in the build.
    dsyer worked on this, see:
    - https://github.com/wro4j/wro4j-gradle-plugin/issues/2
     - https://github.com/dsyer/spring-petclinic/tree/gradle
  changed my mind.  this is not important.  stick with maven for now.
x Replace resilience4j configuration with istio circuit breaker configuration
  x configure the 4s timeout with istio instead of resilience4j.
  x there's also the matter of the fallback when the call times out and its accompanying test.
x Retrofit Visits service so can be configured with a specified delay
x Write up how to test fallback for timeout on visits service endpoint.
x Add version labels to each service
x Upgrade to Spring Boot 3.0
x Question: Has someone already adapted PetClinic for Istio, or otherwise played with Spring Boot on Istio?
  - https://github.com/rverma-jm/spring-petclinic-istio
  - https://piotrminkowski.com/2020/06/01/service-mesh-on-kubernetes-with-istio-and-spring-boot/
  - https://piotrminkowski.com/2020/06/03/circuit-breaker-and-retries-on-kubernetes-with-istio-and-spring-boot/
x Leverage workload identity: add an authorization policy that restricts who can talk to each db
x resolve problem with trace header propagation in boot3 & micrometer
x make project function for either a local or a remote/cloud setup; differences are:
  x ingress endpoint is either localhost or LB_IP of istio-system/istio-ingressgateway
  x push and pull image registries differ

202303..
x give yourself a refresher on the work you did with that app
x review the code, compare with original state
x add to readme a section that discusses the spring cloud gateway replacement
x add to readme a section that discusses spring dependency changes
  x catalog removed dependencies
  x review remaining dependencies:
    boot, actuator, micrometer tracing
    web or reactive web, spring data jpa, mysql jdbc driver/connector
x should be able to access the api endpoints either:
  1. as shown internally from inside the cluster via the sleep pod, for example
  2. via the gateway-exposed routes via the exposed /api/* endpoints
x should mention that the mysql services are also proxied by Envoy.  i actually already do.
x add to readme a section that discusses/documents exposing a prometheus scrape endpoint through "micrometer registry" dependency,
    and how i use the istio conventional annotation to expose app-specific metrics in deployments
x go through readme once more with a remote setup using a gcp-provisioned cluster
x make sure the latest images work (updated spring boot from v3.0.0 to v3.0.5)

2024..
x upgrade spring boot to latest version 3.2.2 and micrometer tracing to 1.2.2
x configure timedAspect per documentation in order to enable @Timed annotation (visits and customers services) for business metrics