test(orcad): skip the live-terminal runtime hand-over across a protocol bump (#24429)

* test(orcad): skip the Bun-to-Node live-terminal hand-over across a protocol bump

The last Bun orcad's daemon reports protocol 38 forever, so asserting the
adopted daemon matches this checkout's PROTOCOL_VERSION failed every bump.
Ask the Bun slot's daemon for its protocol once, run the hand-over when it
matches, and skip with the two versions named when it does not: a daemon
at another protocol is never adopted across an update.

* test(orcad): clean up the Bun protocol probe even when its launch fails

The probe's cleanup ran only after a successful launch, so a launch that
timed out or threw left its orcad and daemon running. One finally now
stops the orcad, kills what it launched, and kills any daemon named by a
pid file in the probe's data root.

.github/workflows/node-server-tests.yml

@@ -125,8 +125,8 @@ jobs:
           pnpm build:orcad-prebuilds --smoke
       - run: pnpm build:orcad
       # Design D7 upgrade and rollback: the last Bun orcad, built from a main commit that shipped
-      # it at this PROTOCOL_VERSION, beside this checkout's Node slot. Same lockfile, so its build
-      # reuses this checkout's node_modules.
+      # it, beside this checkout's Node slot; the live-terminal hand-over skips once PROTOCOL_VERSION
+      # moves past the Bun daemon's. Same lockfile, so its build reuses this checkout's node_modules.
       - uses: oven-sh/setup-bun@0c5077e51419868618aeaa5fe8019c62421857d6 # v2.2.0
         if: runner.os == 'Linux'
         with:

src/main/orcad/orcad-cross-runtime-daemon-adoption.integration.test.ts

@@ -29,6 +29,7 @@ import {
   orcadNodeRuntimeRelativePath
 } from '../../shared/orcad-artifacts'
 import { removeTreeSync } from '../../shared/windows-transient-lock-removal'
+import { readDaemonPidRecord } from '../daemon/daemon-endpoint-incarnation'
 import { PROTOCOL_VERSION } from '../daemon/types'
 import type { ServeReadiness } from '../server/serve-readiness'
 import {
@@ -91,6 +92,7 @@ const missing = [
 let root = ''
 let client = ''
 let backupDriver = ''
+let bunProtocolVersion = 0
 const launched = new Set<number>()
 
 /** Without Vitest's markers: daemon-entry.js does not start its server under VITEST. */
@@ -254,6 +256,40 @@ async function backUpProfile(slot: Slot, runtime: string, userDataDir: string):
   expect(JSON.parse(result.stdout.trim().split('\n').at(-1) ?? '')).toEqual({ ok: true })
 }
 
+function killLaunched(): void {
+  for (const pid of launched) {
+    try {
+      process.kill(pid, 'SIGKILL')
+    } catch {}
+  }
+  launched.clear()
+}
+
+/** The protocol the Bun slot's own daemon reports, read from a throwaway launch. */
+async function probeBunDaemonProtocol(): Promise<number> {
+  const probeRoot = mkdtempSync(join(root, 'bun-probe-'))
+  const slot = installSlot(probeRoot, 'Bun', bunSlotSource!)
+  const userDataDir = join(probeRoot, 'data')
+  try {
+    const daemon = (await launch(slot, userDataDir)).health!.terminalDaemon
+    expect(daemon).toMatchObject({ state: 'live', protocolVersion: expect.any(Number) })
+    return daemon.protocolVersion!
+  } finally {
+    await stop(slot).catch(() => {})
+    killLaunched()
+    // Even after a failed launch: the daemon outlives orcad, and only its pid file names it.
+    const daemonDir = join(userDataDir, 'daemon')
+    for (const name of existsSync(daemonDir) ? readdirSync(daemonDir) : []) {
+      const pid = /^daemon-v\d+\.pid$/.test(name)
+        ? readDaemonPidRecord(join(daemonDir, name))?.pid
+        : undefined
+      if (pid && isAlive(pid)) {
+        process.kill(pid, 'SIGKILL')
+      }
+    }
+  }
+}
+
 // Windows has no orcad launch path (POSIX-only, orcad-remote-host-support.ts), so no inputs.
 const skip = skipForMissingInputs('cross-runtime', posix ? missing : [])
 
@@ -321,27 +357,28 @@ worker.on('error', (error) => { console.error(error); process.exitCode = 1 })
       outfile: client,
       logLevel: 'silent'
     })
-  })
+    bunProtocolVersion = await probeBunDaemonProtocol()
+  }, 180_000)
 
-  afterEach(() => {
-    for (const pid of launched) {
-      try {
-        process.kill(pid, 'SIGKILL')
-      } catch {}
-    }
-    launched.clear()
-  })
+  afterEach(killLaunched)
 
   afterAll(() => {
     removeTreeSync(root)
   })
 
-  it.each([
+  it.for([
     ['Bun', 'Node'],
     ['Node', 'Bun']
   ] as const)(
     '%s orcad hands its daemon and profile to the %s orcad',
-    async (from: Runtime, to: Runtime) => {
+    { timeout: 240_000 },
+    async ([from, to], context) => {
+      // A daemon at another protocol is never adopted across an update, so there is no hand-over.
+      if (bunProtocolVersion !== PROTOCOL_VERSION) {
+        context.skip(
+          `the Bun orcad's daemon speaks protocol ${bunProtocolVersion}, this checkout ${PROTOCOL_VERSION}`
+        )
+      }
       const caseRoot = mkdtempSync(join(root, `${from}-${to}-`))
       const userDataDir = join(caseRoot, 'data')
       const { slots, nodeRuntime } = installSlots(caseRoot)
@@ -426,7 +463,6 @@ worker.on('error', (error) => { console.error(error); process.exitCode = 1 })
           process.kill(daemonPid, 'SIGKILL')
         }
       }
-    },
-    240_000
+    }
   )
 })