From d1bd7b00df6c9e6aa99ef6f16e8081f2722e0dcc Mon Sep 17 00:00:00 2001
From: Teknium <127238744+teknium1@users.noreply.github.com>
Date: Fri, 11 Sep 2026 02:56:15 -0700
Subject: [PATCH] fix(desktop): dispatch probe falls back to /api/status on
 pre-/api/health remotes

Switching the pooled dispatch probe to /api/health (salvaged from #97914)
would 404 on every dispatch against a remote older than 0.19, retire the
tunnel and reconnect forever - the same storm #107997 describes, moved to
old backends. Fall back to /api/status on an explicit 404 exactly the way
the boot readiness probe already does (backend-health.ts). The legacy
fallback idea and its test are taken from #101976 (@edosulai); the rest of
that PR (timeout-tolerance streak, ServerAlive SSH options) is not adopted.

Co-authored-by: Edo Sulaiman <edosulai@icloud.com>
---
 apps/desktop/electron/remote-liveness.test.ts | 30 +++++++++++++++++++
 apps/desktop/electron/remote-liveness.ts      | 21 +++++++++++--
 2 files changed, 48 insertions(+), 3 deletions(-)

diff --git a/apps/desktop/electron/remote-liveness.test.ts b/apps/desktop/electron/remote-liveness.test.ts
index 70e8d45396..fe268bfe60 100644
--- a/apps/desktop/electron/remote-liveness.test.ts
+++ b/apps/desktop/electron/remote-liveness.test.ts
@@ -329,6 +329,36 @@ describe('ensureHealthyPooledRemoteBackendForDispatch', () => {
     expect(retire).toHaveBeenCalledOnce()
     expect(reconnect).toHaveBeenCalledOnce()
   })
+
+  it('falls back to /api/status when /api/health returns 404 on older backends', async () => {
+    const legacy = { baseUrl: 'http://127.0.0.1:49525', mode: 'remote' }
+    const legacyPromise = Promise.resolve(legacy)
+
+    const retire = vi.fn()
+    const reconnect = vi.fn()
+
+    const probe = vi.fn(async (_connection, path) => {
+      if (path === '/api/health') {
+        throw new Error('404: Not Found')
+      }
+    })
+
+    await expect(
+      ensureHealthyPooledRemoteBackendForDispatch({
+        connectionPromise: legacyPromise,
+        currentConnectionPromise: () => legacyPromise,
+        probe,
+        reconnect,
+        retire
+      })
+    ).resolves.toBe(legacy)
+
+    expect(probe).toHaveBeenCalledWith(legacy, '/api/status', {
+      timeoutMs: POOLED_REMOTE_DISPATCH_PROBE_TIMEOUT_MS
+    })
+    expect(retire).not.toHaveBeenCalled()
+    expect(reconnect).not.toHaveBeenCalled()
+  })
 })
 
 describe('revalidatePooledRemoteBackends', () => {
diff --git a/apps/desktop/electron/remote-liveness.ts b/apps/desktop/electron/remote-liveness.ts
index 2160abce6b..f9ab632c6a 100644
--- a/apps/desktop/electron/remote-liveness.ts
+++ b/apps/desktop/electron/remote-liveness.ts
@@ -1,3 +1,5 @@
+import { isMissingHealthEndpointError } from './backend-health'
+
 export const REMOTE_LIVENESS_TIMEOUT_MS = 10_000
 // Dispatch is synchronous user intent: a cached descriptor must prove its
 // forwarded endpoint is alive before it can be returned. Probe cheap
@@ -102,9 +104,22 @@ export async function ensureHealthyPooledRemoteBackendForDispatch<TConnection ex
       return reconnect()
     }
 
-    await probe(connection, '/api/health', {
-      timeoutMs: POOLED_REMOTE_DISPATCH_PROBE_TIMEOUT_MS
-    })
+    try {
+      await probe(connection, '/api/health', {
+        timeoutMs: POOLED_REMOTE_DISPATCH_PROBE_TIMEOUT_MS
+      })
+    } catch (healthError) {
+      // A remote that predates /api/health would otherwise 404 every dispatch,
+      // retire the tunnel and reconnect forever; the boot probe falls back the
+      // same way (backend-health.ts).
+      if (!isMissingHealthEndpointError(healthError)) {
+        throw healthError
+      }
+
+      await probe(connection, '/api/status', {
+        timeoutMs: POOLED_REMOTE_DISPATCH_PROBE_TIMEOUT_MS
+      })
+    }
   } catch (error) {
     if (currentConnectionPromise() === connectionPromise) {
       await retire(error)