Skip to content

Latest commit

 

History

History
75 lines (51 loc) · 2.24 KB

TODO.md

File metadata and controls

75 lines (51 loc) · 2.24 KB

Tier 1

  • Tier1 GET should return list of cloudlets without deploying

  • Better documentation for cloudlet.yaml local configuration file

  • Improve error reporting on failed deployments, right now it is just a 500 error

Tier 2

  • figure out how to discover when the kubernetes cluster is actually fully deployed and ready to deploy backends. Althought the tier2 pod starts pretty quickly, it takes almost 4 minutes before all the other pods are done after a reboot. And this is on a cloudlet where all containers are fully cached and kernel modules have already been built. Right now we are already reporting to tier1 before we're actually ready to accept work.

  • How to handle a multi-node cloudlet, we have to check resource utilization on a per node basis and report the best available one(s)?

  • Tier2 to Tier1 reporting

    %: v cpu v memory v uplink v downlink v gpu utilization o loadavg

  • report USE metrics (usage, saturation, errors)?

  • expand reporting to Tier1 to include

    • location
    • accepted_clients
    • rejected_clients
    • local_networks?

  • clean up connecting to cluster, right now there is an ugly hack to fix --help when we don't have cluster credentials.

  • merge Tier2/cluster.py with Tier1/cloudlets.py, useful to track common items to report like location and client ACLs

  • handle prometheus port tunnelling for non-local cloudlets

    Check for the KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT environment variables and the existence of service account token file at /var/run/secrets/kubernetes.io/serviceaccount/token.

    maybe we can leverage the existing Kilo VPN infrastructure?

  • report known/supported application uuids to Tier 1?

  • Improve error reporting on failed deployments, right now it is just a 500 error

Tier 3