Backport of client: ignore restart issued to terminal allocations into release/1.5.x (#17211)

This pull request was automerged via backport-assistant

Author: hc-github-team-nomad-core

.changelog/17175.txt

@@ -0,0 +1,3 @@
+```release-note:bug
+client: Fixed a bug where restarting a terminal allocation turns it into a zombie where allocation and task hooks will run unexpectedly
+```

client/allocrunner/alloc_runner.go

@@ -1264,6 +1264,12 @@ func (ar *allocRunner) RestartAll(event *structs.TaskEvent) error {
 
 // restartTasks restarts all task runners concurrently.
 func (ar *allocRunner) restartTasks(ctx context.Context, event *structs.TaskEvent, failure bool, force bool) error {
+
+	// ensure we are not trying to restart an alloc that is terminal
+	if !ar.shouldRun() {
+		return fmt.Errorf("restart of an alloc that should not run")
+	}
+
 	waitCh := make(chan struct{})
 	var err *multierror.Error
 	var errMutex sync.Mutex

e2e/clientstate/allocs_test.go

@@ -0,0 +1,61 @@
+// Copyright (c) HashiCorp, Inc.
+// SPDX-License-Identifier: MPL-2.0
+
+package clientstate
+
+import (
+	"testing"
+	"time"
+
+	"github.com/hashicorp/nomad/e2e/e2eutil"
+	"github.com/hashicorp/nomad/helper/uuid"
+	"github.com/shoenig/test/must"
+	"github.com/shoenig/test/wait"
+)
+
+func TestClientAllocs(t *testing.T) {
+	nomad := e2eutil.NomadClient(t)
+
+	e2eutil.WaitForLeader(t, nomad)
+	e2eutil.WaitForNodesReady(t, nomad, 1)
+
+	t.Run("testAllocZombie", testAllocZombie)
+}
+
+// testAllocZombie ensures that a restart of a dead allocation does not cause
+// it to come back to life in a not-quite alive state.
+//
+// https://github.com/hashicorp/nomad/issues/17079
+func testAllocZombie(t *testing.T) {
+	nomad := e2eutil.NomadClient(t)
+
+	jobID := "alloc-zombie-" + uuid.Short()
+	jobIDs := []string{jobID}
+	t.Cleanup(e2eutil.CleanupJobsAndGC(t, &jobIDs))
+
+	// start the job and wait for alloc to become failed
+	err := e2eutil.Register(jobID, "./input/alloc_zombie.nomad")
+	must.NoError(t, err)
+
+	allocID := e2eutil.SingleAllocID(t, jobID, "", 0)
+
+	// wait for alloc to be marked as failed
+	e2eutil.WaitForAllocStatus(t, nomad, allocID, "failed")
+
+	// wait for additional failures to know we got rescheduled
+	must.Wait(t, wait.InitialSuccess(
+		wait.BoolFunc(func() bool {
+			statuses, err := e2eutil.AllocStatusesRescheduled(jobID, "")
+			must.NoError(t, err)
+			return len(statuses) > 2
+		}),
+		wait.Timeout(1*time.Minute),
+		wait.Gap(1*time.Second),
+	))
+
+	// now attempt to restart our initial allocation
+	// which should do nothing but give us an error
+	output, err := e2eutil.Command("nomad", "alloc", "restart", allocID)
+	must.ErrorContains(t, err, "restart of an alloc that should not run")
+	must.StrContains(t, output, "Failed to restart allocation")
+}

e2e/clientstate/input/alloc_zombie.nomad

@@ -0,0 +1,59 @@
+# Copyright (c) HashiCorp, Inc.
+# SPDX-License-Identifier: MPL-2.0
+
+job "alloc_zombie" {
+
+  group "group" {
+    network {
+      mode = "host"
+      port "http" {}
+    }
+
+    service {
+      name     = "alloczombie"
+      port     = "http"
+      provider = "nomad"
+
+      check {
+        name     = "alloczombiecheck"
+        type     = "http"
+        port     = "http"
+        path     = "/does/not/exist.txt"
+        interval = "2s"
+        timeout  = "1s"
+        check_restart {
+          limit = 1
+          grace = "3s"
+        }
+      }
+    }
+
+    reschedule {
+      attempts       = 3
+      interval       = "1m"
+      delay          = "5s"
+      delay_function = "constant"
+      unlimited      = false
+    }
+
+    restart {
+      attempts = 0
+      delay    = "5s"
+      mode     = "fail"
+    }
+
+    task "python" {
+      driver = "raw_exec"
+
+      config {
+        command = "python3"
+        args    = ["-m", "http.server", "${NOMAD_PORT_http}", "--directory", "/tmp"]
+      }
+
+      resources {
+        cpu    = 10
+        memory = 64
+      }
+    }
+  }
+}