mirror of
https://github.com/mulgadc/spinifex.git
synced 2026-10-19 19:01:42 +00:00
Josh Siviour
92fa467c90
* refactor(e2e/multinode): flatten phase 11 + parallelize bucket #1 + drop phase numbers Split monolithic phase11_PlacementAndNATGateway (~430 lines, 8 inline sub-steps) into 6 top-level Tests, each owning its own VPC graph via t.Cleanup and layered setup helpers (spreadSetupVPC, spreadSetupBastion, spreadSetupPrivateTrio, spreadSetupNATGW). The cleanup-ordering test flips *bool latches as each explicit teardown step succeeds so the LIFO unwind is a no-op on the success path. Add t.Parallel() to read-only / independent Tests: ClusterHealth, InstanceDistribution, GuestSSH, VPCNetworking. Deviates from the mulga-siv-127 bead's bucket-#1 listing by keeping CrossNodeGateway sequential (asserts a snapshot equality across gateways that the parallel VPC test would break by churning instance count mid-assert) and CrossNodeOps sequential (stops/starts trio[0], would race GuestSSH). Both deviations documented in tests_test.go + plan doc. Drop the bash phase numbering (phaseN_X funcs + "Phase N" log strings) from the rest of the suite — 10 funcs renamed to runX. Phase numbers were a port-time crutch; the Test names now describe each scenario and phase numbers ageing into noise as bash is retired in Stage H. * refactor(e2e/multinode): collapse spread Tests + control execution order Two changes wedged together since they touch the same files: 1. Collapse the 6 top-level TestMultinodeSpread* Tests into a single TestMultinodeSpread with 6 t.Run sub-tests sharing one VPC + bastion + private trio + NAT GW setup chain. The 6-test split paid setup cost 6× — bastion launch + private trio launch + NAT GW provision per Test pushed phase-11 wall-clock near 60min vs ~15min monolithic. Shared- setup layout keeps JUnit granularity (one row per sub-test) while restoring the monolithic budget. Setup helpers register t.Cleanup against the outer test so teardown runs once at the end (LIFO: NAT GW → trio → bastion → VPC). 2. Move the TestMultinodeSpread wrapper into tests_test.go so file-source order controls execution position. Previously the func lived in placement_nat_test.go which sorts before preflight_test.go and tests_test.go alphabetically — Spread ran FIRST in the sequential bucket, defeating Preflight's fail-fast on a sick cluster. Wrapper now runs after NodeRecovery so the cluster has been fully degrade-tested before Spread's 4-VM + NAT GW + custom-VPC graph spins up.
45 lines
1.3 KiB
Go
45 lines
1.3 KiB
Go
//go:build e2e
|
|
|
|
package multinode
|
|
|
|
import (
|
|
"context"
|
|
"os"
|
|
"testing"
|
|
"time"
|
|
|
|
"github.com/mulgadc/spinifex/tests/e2e/harness"
|
|
)
|
|
|
|
// runPreflight is the Go port of pre-flight validation
|
|
// (run-multinode-e2e.sh:394-423). Two checks:
|
|
// - /dev/kvm exists and is writable on the local node (we run on node1).
|
|
// - SSH to every peer node succeeds with `hostname`.
|
|
//
|
|
// Failing this exposes a misconfigured VM image or broken peer networking
|
|
// before any AWS API churn.
|
|
func runPreflight(t *testing.T, fix *Fixture) {
|
|
harness.Phase(t, "Multinode — Pre-flight")
|
|
|
|
harness.Step(t, "verify /dev/kvm")
|
|
f, err := os.OpenFile("/dev/kvm", os.O_RDWR, 0)
|
|
if err != nil {
|
|
t.Fatalf("/dev/kvm not writable: %v", err)
|
|
}
|
|
_ = f.Close()
|
|
|
|
harness.Step(t, "peer_ssh hostname on every remote node")
|
|
ssh := harness.NewPeerSSH()
|
|
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
|
|
defer cancel()
|
|
// Bash runs this loop from node1 and skips itself; mirror that — the
|
|
// runner host doesn't necessarily accept SSH back into its own loopback
|
|
// and a local hostname call would test nothing about peer reachability.
|
|
for _, n := range fix.Cluster.Nodes[1:] {
|
|
if _, err := ssh.Run(ctx, n.Addr, "hostname"); err != nil {
|
|
t.Fatalf("peer_ssh %s (%s): %v", n.Name, n.Addr, err)
|
|
}
|
|
harness.Detail(t, "node", n.Name, "addr", n.Addr, "ssh", "ok")
|
|
}
|
|
}
|