Skip to content

Latest commit

 

History

History
37 lines (25 loc) · 1.75 KB

File metadata and controls

37 lines (25 loc) · 1.75 KB

Phase 11 — Network negative verification

Phase 11 hardens the virtual mesh lab against network failure modes that happy-path verify scripts do not cover. The director control plane (peer-a) is tested separately from worker reachability on the Docker bridge.

Depends on: peer-a + peer-b running, Phase 9 LiteDB on peer-a (for director-restart lease test).

# Scenario Expect
1 Fleet peer with blackhole apiBaseUrl (10.255.255.254) Schedule → Assigned; bridge curl to worker fails (no preflight)
2 Fleet peer with unresolvable DNS (.invalid host) Schedule → Assigned; bridge curl fails
3 Drained-only fleet Schedule → 400; placementReason indicates no eligible nodes
4 peer-b stopped while task Running Bridge cannot reach peer-b:8080; after start, health OK; director PATCH still works
5 peer-a restart (LiteDB) Task + leaseToken survive; PATCH Running succeeds

Shared helpers: scripts/mesh-lab-net.sh.

Run

make mesh-lab-e2e-workers          # includes network-negative after trust
make mesh-lab-verify-network-negative   # lab already up

Skip: MESH_LAB_SKIP_NETWORK_NEGATIVE_VERIFY=1

What this does not replace

Revision history

Date Change
2026-05-19 Initial network-negative verify script + CI wiring.