Meeting 2017 06 23

• Time: 16:15 - 17:30

• Participants: Tim, Matthias, Lennart

• Tim

  • (Conference in between, thus other work)
  • Dovecot with mirrored GlusterFS between EU and US
    • This GlusterFS can be mounted by all relevant pods in k8s, nice
    • File system will still need to get encrypted, but mostly working

• Matthias

  • Prometheus now scrapes our GCE-deployed pods, some regex magic was needed
  • Grafana deployed, visualization of all metrics collected from k8s
  • Looked at Dovecot exporter to Prometheus, could get tricky

• Lennart

  • Finally gRPC integrated in pluto, nice
  • Reviewed Tim's IMAP benchmark
  • Checked IMAP CLOSE command, we will probably not need it
  • Started to review Isabelle/HOL proof comments

• TO

  • Evaluation
    • 1) pluto, Dovecot 1 (naïve), Dovecot 2 (#smart), Gmail: response time
    • 2) pluto, Dovecot 1 (naïve), Dovecot 2 (#smart), Gmail: APPEND message synchronization time around the globe
    • 3) pluto in depth: multi-cluster around the globe, very deep internal statistics, exact internal synchronization

• ToDos until 2017-06-29

  • Matthias:
    • Dovecot exporter?
    • Move Pluto into own namespace
    • gokit level.go:63 default caller problem
    • Fail of make docker build pipeline step
    • Deploy final pluto setup on k8s
    • Recreate Kubernetes cluster with bigger nodes
    • Have a look which and where to export Prometheus metrics from pluto
    • Start styx?
  • Lennart:
    • Implement gRPC-based worker => storage failover
    • Prometheus metrics in pluto
    • Finish gRPC-adapted tests?
    • Maildir Prometheus exporter
    • IMAP benchmark continuation
  • Tim:
    • IMAP benchmark continuation
    • Diverse Dovecot k8s deployments
    • TODO