prisma-risk / tsoracle / 26388776776
95%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 64
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 95.644%. Remained the same
26388776776

push

github

web-flow 
test(e2e): sketch a kind-based multi-node cluster harness (#405)

Adds an opt-in Kubernetes (kind) end-to-end lane that runs a 3-node
openraft-standalone tsoracle cluster, plus the design doc that scopes it.
This is scaffolding for review only — nothing is wired into CI.

The existing harnesses cover protocol correctness deterministically but
never exercise the deployment envelope: the process topology is
single-node, and the raft/paxos topologies share an in-process MemNetwork
with no sockets, DNS, or real peer transport. This lane fills exactly that
envelope (cluster formation over real gRPC, StatefulSet identity, PVC
reattach, lifecycle events) and nothing below it.

Contents:
- docs/kubernetes-e2e.md: gap analysis, scope, architecture, tradeoffs,
  and a staged rollout plan.
- e2e/kube/: multi-stage Dockerfile for the openraft-standalone binary, an
  ordinal-derived entrypoint, a kind cluster config, and StatefulSet +
  headless/client Services + PodDisruptionBudget manifests.

Readiness is a TCP check on the tso port, not a GetTs probe: a follower
returning a leader-hint redirect is healthy and must stay in rotation.

Known gap called out in the doc and left for a follow-up PR: the
standalone example binaries handle only SIGINT, not SIGTERM, so pod drains
ride out the grace period and end in SIGKILL. The stock tsoracle serve
binary already handles SIGTERM (#245); the examples a cluster runs do not.

9309 of 9733 relevant lines covered (95.64%)

587802.32 hits per line

Jobs
ID Job ID Ran Files Coverage
1 26388776776.1 64
95.64
 GitHub Action Run
Source Files on build 26388776776