backport of commit a1299ad

hashicorp · Oct 18, 2022 · 7a5c68b · 7a5c68b
1 parent 1b43418
commit 7a5c68b
Show file tree

Hide file tree

Showing 86 changed files with 685 additions and 2,113 deletions.
diff --git a/.changelog/14492.txt b/.changelog/14492.txt
diff --git a/.changelog/14775.txt b/.changelog/14775.txt
diff --git a/.changelog/14917.txt b/.changelog/14917.txt
@@ -0,0 +1,3 @@
+```release-note:bug
+consul: Fixed a bug where services continuously re-registered
+```
diff --git a/.changelog/14924.txt b/.changelog/14924.txt
@@ -0,0 +1,4 @@
+```release-note:bug
+nomad native service discovery: Ensure all local servers meet v.1.3.0 minimum before service registrations can be written
+```
+
diff --git a/.github/workflows/lock.yml b/.github/workflows/lock.yml
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,16 +1,20 @@
+## 1.4.1 (October 06, 2022)
+
+BUG FIXES:
+
+* keyring: Fixed a panic that can occur during upgrades to 1.4.0 when initializing the keyring [[GH-14821](https://github.com/hashicorp/nomad/issues/14821)]
+
 ## 1.4.0 (October 04, 2022)
 
 FEATURES:
 
 * **ACL Roles:** Added support for ACL Roles. [[GH-14320](https://github.com/hashicorp/nomad/issues/14320)]
-* **Nomad Native Service Discovery**: Add built-in support for checks on Nomad services [[GH-13715](https://github.com/hashicorp/nomad/issues/13715)]
 * **Variables:** Added support for storing encrypted configuration values. [[GH-13000](https://github.com/hashicorp/nomad/issues/13000)]
-* **UI Services table:** Display task-level services in addition to group-level services. [[GH-14199](https://github.com/hashicorp/nomad/issues/14199)]
+* [ui] Services table: Display task-level services in addition to group-level services. [[GH-14199](https://github.com/hashicorp/nomad/issues/14199)]
 
 BREAKING CHANGES:
 
 * audit (Enterprise): fixed inconsistency in event filter logic [[GH-14212](https://github.com/hashicorp/nomad/issues/14212)]
-* cli: `eval status -json` no longer supports listing all evals in JSON. Use `eval list -json`. [[GH-14651](https://github.com/hashicorp/nomad/issues/14651)]
 * core: remove support for raft protocol version 2 [[GH-13467](https://github.com/hashicorp/nomad/issues/13467)]
 
 SECURITY:
@@ -22,6 +26,8 @@ IMPROVEMENTS:
 * acl: ACL tokens can now be created with an expiration TTL. [[GH-14320](https://github.com/hashicorp/nomad/issues/14320)]
 * api: return a more descriptive error when /v1/acl/bootstrap fails to decode request body [[GH-14629](https://github.com/hashicorp/nomad/issues/14629)]
 * autopilot: upgrade to raft-autopilot library [[GH-14441](https://github.com/hashicorp/nomad/issues/14441)]
+* build: Update go toolchain to 1.18.5 [[GH-13539](https://github.com/hashicorp/nomad/issues/13539)]
+* build: update Go to version go1.18.2 [[GH-13036](https://github.com/hashicorp/nomad/issues/13036)]
 * cli: Removed deprecated network quota fields from `quota status` output [[GH-14468](https://github.com/hashicorp/nomad/issues/14468)]
 * cli: `acl policy info` output format has changed to improve readability with large policy documents [[GH-14140](https://github.com/hashicorp/nomad/issues/14140)]
 * cli: `operator debug` now writes newline-delimited JSON files for large collections [[GH-14610](https://github.com/hashicorp/nomad/issues/14610)]
@@ -47,44 +53,15 @@ IMPROVEMENTS:
 * server: Log when a node's eligibility changes [[GH-14125](https://github.com/hashicorp/nomad/issues/14125)]
 * ui: Display different message when trying to exec into a job with no task running. [[GH-14071](https://github.com/hashicorp/nomad/issues/14071)]
 * ui: add service discovery, along with health checks, to job and allocation routes [[GH-14408](https://github.com/hashicorp/nomad/issues/14408)]
+* ui: added visual regression tests for top-level UI routes [[GH-12872](https://github.com/hashicorp/nomad/issues/12872)]
 * ui: adds a sidebar to show in-page logs for a given task, accessible via job, client, or task group routes [[GH-14612](https://github.com/hashicorp/nomad/issues/14612)]
 * ui: allow deep-dive clicks to tasks from client, job, and task group routes. [[GH-14592](https://github.com/hashicorp/nomad/issues/14592)]
 * ui: attach timestamps and a visual indicator on failure to health checks in the Web UI [[GH-14677](https://github.com/hashicorp/nomad/issues/14677)]
+* ui: warn a user before they leave a Variables form page with unsaved information [[GH-14665](https://github.com/hashicorp/nomad/issues/14665)]
 
-BUG FIXES:
-
-* api: Fixed a bug where the List Volume API did not include the `ControllerRequired` and `ResourceExhausted` fields. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* cli: Ignore Vault token when generating job diff. [[GH-14424](https://github.com/hashicorp/nomad/issues/14424)]
-* cli: fixed a bug in the `operator api` command where the HTTPS scheme was not always correctly calculated [[GH-14635](https://github.com/hashicorp/nomad/issues/14635)]
-* cli: return exit code `255` when `nomad job plan` fails job validation. [[GH-14426](https://github.com/hashicorp/nomad/issues/14426)]
-* cli: set content length on POST requests when using the `nomad operator api` command [[GH-14634](https://github.com/hashicorp/nomad/issues/14634)]
-* client: Fixed bug where clients could attempt to connect to servers with invalid addresses retrieved from Consul. [[GH-14431](https://github.com/hashicorp/nomad/issues/14431)]
-* core: prevent new allocations from overlapping execution with stopping allocations [[GH-10446](https://github.com/hashicorp/nomad/issues/10446)]
-* csi: Fixed a bug where a volume that was successfully unmounted by the client but then failed controller unpublishing would not be marked free until garbage collection ran. [[GH-14675](https://github.com/hashicorp/nomad/issues/14675)]
-* csi: Fixed a bug where the server would not send controller unpublish for a failed allocation. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* csi: Fixed a data race in the volume unpublish endpoint that could result in claims being incorrectly marked as freed before being persisted to raft. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* helpers: Fixed a bug where random stagger func did not protect against negative inputs [[GH-14497](https://github.com/hashicorp/nomad/issues/14497)]
-* jobspec: Fixed a bug where an `artifact` with `headers` configuration would fail to parse when using HCLv1 [[GH-14637](https://github.com/hashicorp/nomad/issues/14637)]
-* metrics: Update client `node_scheduling_eligibility` value with server heartbeats. [[GH-14483](https://github.com/hashicorp/nomad/issues/14483)]
-* quotas (Enterprise): Fixed a server crashing panic when updating and checking a quota concurrently.
-* rpc (Enterprise): check for spec changes in all regions when registering multiregion jobs [[GH-14519](https://github.com/hashicorp/nomad/issues/14519)]
-* scheduler (Enterprise): Fixed bug where the scheduler would treat multiregion jobs as paused for job types that don't use deployments [[GH-14659](https://github.com/hashicorp/nomad/issues/14659)]
-* template: Fixed a bug where the `splay` timeout was not being applied when `change_mode` was set to `script`. [[GH-14749](https://github.com/hashicorp/nomad/issues/14749)]
-* ui: Remove extra space when displaying the version in the menu footer. [[GH-14457](https://github.com/hashicorp/nomad/issues/14457)]
-
-## 1.3.6 (October 04, 2022)
-
-SECURITY:
-
-* client: recover from panics caused by artifact download to prevent the Nomad client from crashing [[GH-14696](https://github.com/hashicorp/nomad/issues/14696)]
-
-IMPROVEMENTS:
+DEPRECATIONS:
 
-* api: return a more descriptive error when /v1/acl/bootstrap fails to decode request body [[GH-14629](https://github.com/hashicorp/nomad/issues/14629)]
-* cli: ignore `-hcl2-strict` when -hcl1 is set. [[GH-14426](https://github.com/hashicorp/nomad/issues/14426)]
-* cli: warn destructive update only when count is greater than 1 [[GH-13103](https://github.com/hashicorp/nomad/issues/13103)]
-* consul: Allow interpolation of task environment values into Consul Service Mesh configuration [[GH-14445](https://github.com/hashicorp/nomad/issues/14445)]
-* ui: Display different message when trying to exec into a job with no task running. [[GH-14071](https://github.com/hashicorp/nomad/issues/14071)]
+* cli: `eval status -json` no longer supports listing all evals in JSON. Use `eval list -json`. [[GH-14651](https://github.com/hashicorp/nomad/issues/14651)]
 
 BUG FIXES:
 
@@ -94,6 +71,7 @@ BUG FIXES:
 * cli: return exit code `255` when `nomad job plan` fails job validation. [[GH-14426](https://github.com/hashicorp/nomad/issues/14426)]
 * cli: set content length on POST requests when using the `nomad operator api` command [[GH-14634](https://github.com/hashicorp/nomad/issues/14634)]
 * client: Fixed bug where clients could attempt to connect to servers with invalid addresses retrieved from Consul. [[GH-14431](https://github.com/hashicorp/nomad/issues/14431)]
+* core: prevent new allocations from overlapping execution with stopping allocations [[GH-10446](https://github.com/hashicorp/nomad/issues/10446)]
 * csi: Fixed a bug where a volume that was successfully unmounted by the client but then failed controller unpublishing would not be marked free until garbage collection ran. [[GH-14675](https://github.com/hashicorp/nomad/issues/14675)]
 * csi: Fixed a bug where the server would not send controller unpublish for a failed allocation. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
 * csi: Fixed a data race in the volume unpublish endpoint that could result in claims being incorrectly marked as freed before being persisted to raft. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
@@ -105,6 +83,7 @@ BUG FIXES:
 * scheduler: Fixed bug where the scheduler would treat multiregion jobs as paused for job types that don't use deployments [[GH-14659](https://github.com/hashicorp/nomad/issues/14659)]
 * template: Fixed a bug where the `splay` timeout was not being applied when `change_mode` was set to `script`. [[GH-14749](https://github.com/hashicorp/nomad/issues/14749)]
 * ui: Remove extra space when displaying the version in the menu footer. [[GH-14457](https://github.com/hashicorp/nomad/issues/14457)]
+* ui: Stabilizes visual regression tests [[GH-14551](https://github.com/hashicorp/nomad/issues/14551)]
 
 ## 1.3.5 (August 31, 2022)
 
@@ -382,25 +361,6 @@ BUG FIXES:
 * ui: fix broken link to task-groups in the Recent Allocations table in the Job Detail overview page. [[GH-12765](https://github.com/hashicorp/nomad/issues/12765)]
 * ui: fix the unit for the task row memory usage metric [[GH-11980](https://github.com/hashicorp/nomad/issues/11980)]
 
-## 1.2.13 (October 04, 2022)
-
-SECURITY:
-
-* client: recover from panics caused by artifact download to prevent the Nomad client from crashing [[GH-14696](https://github.com/hashicorp/nomad/issues/14696)]
-
-BUG FIXES:
-
-* api: Fixed a bug where the List Volume API did not include the `ControllerRequired` and `ResourceExhausted` fields. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* client: Fixed bug where clients could attempt to connect to servers with invalid addresses retrieved from Consul. [[GH-14431](https://github.com/hashicorp/nomad/issues/14431)]
-* csi: Fixed a bug where a volume that was successfully unmounted by the client but then failed controller unpublishing would not be marked free until garbage collection ran. [[GH-14675](https://github.com/hashicorp/nomad/issues/14675)]
-* csi: Fixed a bug where the server would not send controller unpublish for a failed allocation. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* csi: Fixed a bug where volume claims on lost or garbage collected nodes could not be freed [[GH-14720](https://github.com/hashicorp/nomad/issues/14720)]
-* csi: Fixed a data race in the volume unpublish endpoint that could result in claims being incorrectly marked as freed before being persisted to raft. [[GH-14484](https://github.com/hashicorp/nomad/issues/14484)]
-* jobspec: Fixed a bug where an `artifact` with `headers` configuration would fail to parse when using HCLv1 [[GH-14637](https://github.com/hashicorp/nomad/issues/14637)]
-* metrics: Update client `node_scheduling_eligibility` value with server heartbeats. [[GH-14483](https://github.com/hashicorp/nomad/issues/14483)]
-* quotas (Enterprise): Fixed a server crashing panic when updating and checking a quota concurrently.
-* rpc: check for spec changes in all regions when registering multiregion jobs [[GH-14519](https://github.com/hashicorp/nomad/issues/14519)]
-
 ## 1.2.12 (August 31, 2022)
 
 IMPROVEMENTS:

diff --git a/GNUmakefile b/GNUmakefile
@@ -49,7 +49,7 @@ PROTO_COMPARE_TAG ?= v1.0.3$(if $(findstring ent,$(GO_TAGS)),+ent,)
 
 # LAST_RELEASE is the git sha of the latest release corresponding to this branch. main should have the latest
 # published release, and release branches should point to the latest published release in the X.Y release line.
-LAST_RELEASE ?= v1.4.0
+LAST_RELEASE ?= v1.4.1
 
 default: help
 

diff --git a/LICENSE b/LICENSE
@@ -1,5 +1,3 @@
-Copyright (c) 2015 HashiCorp, Inc.
-
 Mozilla Public License, version 2.0
 
 1. Definitions

diff --git a/client/allocdir/fs_unix.go b/client/allocdir/fs_unix.go
@@ -1,4 +1,4 @@
-//go:build unix
+//go:build darwin || dragonfly || freebsd || linux || netbsd || openbsd || solaris
 
 package allocdir
 
@@ -40,14 +40,17 @@ func dropDirPermissions(path string, desired os.FileMode) error {
 		return nil
 	}
 
-	nobody := users.Nobody()
+	nobody, err := users.Nobody()
+	if err != nil {
+		return err
+	}
 
-	uid, err := getUid(&nobody)
+	uid, err := getUid(nobody)
 	if err != nil {
 		return err
 	}
 
-	gid, err := getGid(&nobody)
+	gid, err := getGid(nobody)
 	if err != nil {
 		return err
 	}

diff --git a/client/allocrunner/taskrunner/task_runner.go b/client/allocrunner/taskrunner/task_runner.go
@@ -982,10 +982,6 @@ func (tr *TaskRunner) handleKill(resultCh <-chan *drivers.ExitResult) *drivers.E
 	if delay := tr.Task().ShutdownDelay; delay != 0 {
 		tr.logger.Debug("waiting before killing task", "shutdown_delay", delay)
 
-		ev := structs.NewTaskEvent(structs.TaskWaitingShuttingDownDelay).
-			SetDisplayMessage(fmt.Sprintf("Waiting for shutdown_delay of %s before killing the task.", delay))
-		tr.UpdateState(structs.TaskStatePending, ev)
-
 		select {
 		case result := <-resultCh:
 			return result

diff --git a/client/allocrunner/taskrunner/template/template_test.go b/client/allocrunner/taskrunner/template/template_test.go
@@ -24,7 +24,6 @@ import (
 	"github.com/hashicorp/nomad/client/allocdir"
 	"github.com/hashicorp/nomad/client/config"
 	"github.com/hashicorp/nomad/client/taskenv"
-	clienttestutil "github.com/hashicorp/nomad/client/testutil"
 	"github.com/hashicorp/nomad/helper/pointer"
 	"github.com/hashicorp/nomad/helper/testlog"
 	"github.com/hashicorp/nomad/helper/users"
@@ -185,7 +184,7 @@ func newTestHarness(t *testing.T, templates []*structs.Template, consul, vault b
 	if consul {
 		var err error
 		harness.consul, err = ctestutil.NewTestServerConfigT(t, func(c *ctestutil.TestServerConfig) {
-			c.Peering = nil // fix for older versions of Consul (<1.13.0) that don't support peering
+			c.Peering = nil  // fix for older versions of Consul (<1.13.0) that don't support peering
 		})
 		if err != nil {
 			t.Fatalf("error starting test Consul server: %v", err)
@@ -514,7 +513,6 @@ func TestTaskTemplateManager_Unblock_Static(t *testing.T) {
 }
 
 func TestTaskTemplateManager_Permissions(t *testing.T) {
-	clienttestutil.RequireRoot(t)
 	ci.Parallel(t)
 	// Make a template that will render immediately
 	content := "hello, world!"

diff --git a/client/client.go b/client/client.go
@@ -1571,7 +1571,7 @@ func (c *Client) updateNodeFromFingerprint(response *fingerprint.FingerprintResp
 		response.Resources.Networks = updateNetworks(
 			response.Resources.Networks,
 			newConfig)
-		if !newConfig.Node.Resources.Equal(response.Resources) {
+		if !newConfig.Node.Resources.Equals(response.Resources) {
 			newConfig.Node.Resources.Merge(response.Resources)
 			nodeHasChanged = true
 		}
@@ -1583,7 +1583,7 @@ func (c *Client) updateNodeFromFingerprint(response *fingerprint.FingerprintResp
 		response.NodeResources.Networks = updateNetworks(
 			response.NodeResources.Networks,
 			newConfig)
-		if !newConfig.Node.NodeResources.Equal(response.NodeResources) {
+		if !newConfig.Node.NodeResources.Equals(response.NodeResources) {
 			newConfig.Node.NodeResources.Merge(response.NodeResources)
 			nodeHasChanged = true
 		}

diff --git a/client/config/config.go b/client/config/config.go
@@ -3,6 +3,8 @@ package config
 import (
 	"errors"
 	"fmt"
+	"io"
+	"os"
 	"reflect"
 	"strconv"
 	"strings"
@@ -46,7 +48,7 @@ var (
 		"java",
 	}, ",")
 
-	// DefaultChrootEnv is a mapping of directories on the host OS to attempt to embed inside each
+	// A mapping of directories on the host OS to attempt to embed inside each
 	// task's chroot.
 	DefaultChrootEnv = map[string]string{
 		"/bin":            "/bin",
@@ -92,6 +94,9 @@ type Config struct {
 	// AllocDir is where we store data for allocations
 	AllocDir string
 
+	// LogOutput is the destination for logs
+	LogOutput io.Writer
+
 	// Logger provides a logger to the client
 	Logger log.InterceptLogger
 
@@ -196,6 +201,9 @@ type Config struct {
 	// before garbage collection is triggered.
 	GCMaxAllocs int
 
+	// LogLevel is the level of the logs to putout
+	LogLevel string
+
 	// NoHostUUID disables using the host's UUID and will force generation of a
 	// random UUID.
 	NoHostUUID bool
@@ -453,8 +461,8 @@ func (wc *WaitConfig) Copy() *WaitConfig {
 	return wc
 }
 
-// Equal returns the result of reflect.DeepEqual
-func (wc *WaitConfig) Equal(other *WaitConfig) bool {
+// Equals returns the result of reflect.DeepEqual
+func (wc *WaitConfig) Equals(other *WaitConfig) bool {
 	return reflect.DeepEqual(wc, other)
 }
 
@@ -463,7 +471,7 @@ func (wc *WaitConfig) IsEmpty() bool {
 	if wc == nil {
 		return true
 	}
-	return wc.Equal(&WaitConfig{})
+	return wc.Equals(&WaitConfig{})
 }
 
 // Validate returns an error  if the receiver is nil or empty or if Min is greater
@@ -588,8 +596,8 @@ func (rc *RetryConfig) Copy() *RetryConfig {
 	return nrc
 }
 
-// Equal returns the result of reflect.DeepEqual
-func (rc *RetryConfig) Equal(other *RetryConfig) bool {
+// Equals returns the result of reflect.DeepEqual
+func (rc *RetryConfig) Equals(other *RetryConfig) bool {
 	return reflect.DeepEqual(rc, other)
 }
 
@@ -599,7 +607,7 @@ func (rc *RetryConfig) IsEmpty() bool {
 		return true
 	}
 
-	return rc.Equal(&RetryConfig{})
+	return rc.Equals(&RetryConfig{})
 }
 
 // Validate returns an error if the receiver is nil or empty, or if Backoff
@@ -714,9 +722,11 @@ func DefaultConfig() *Config {
 		Version:                 version.GetVersion(),
 		VaultConfig:             structsc.DefaultVaultConfig(),
 		ConsulConfig:            structsc.DefaultConsulConfig(),
+		LogOutput:               os.Stderr,
 		Region:                  "global",
 		StatsCollectionInterval: 1 * time.Second,
 		TLSConfig:               &structsc.TLSConfig{},
+		LogLevel:                "DEBUG",
 		GCInterval:              1 * time.Minute,
 		GCParallelDestroys:      2,
 		GCDiskUsageThreshold:    80,