kemko/nomad
1
0
Fork 0
mirror of https://github.com/kemko/nomad.git synced 2026-01-08 19:35:41 +03:00
Code Issues Packages Projects Releases Wiki Activity

Fixed merge conflicts

Browse Source
This commit is contained in:
Diptanu Choudhury
2016-02-08 10:24:27 -08:00
parent 4a01bd8bcd d7a772c52b
commit 9a6da514e1
39 changed files with 573 additions and 166 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.travis.yml
View File

@@ -11,7 +11,7 @@ go:
- tip
env:
- TRAVIS_RUN=true DOCKER_VERSION=1.9.1
- DOCKER_VERSION=1.9.1
matrix:
allow_failures:

5
CHANGELOG.md
View File

@@ -3,6 +3,7 @@
BACKWARDS INCOMPATIBILITIES:
* core: Improved restart policy with more user configuration [GH-594]
* core/cli: Print short identifiers [GH-675]
* core/consul: Validate service name doesn't include period [GH-770]
* core/jobspec: Variables/constraints interpreted using ${} notation [GH-675]
* client: Environment variable containing address for each allocated port
[GH-704]
@@ -26,6 +27,7 @@ IMPROVEMENTS:
* core/client: Client pulls minimum set of required allocations [GH-731]
* core/jobspec: Default task resources and validation [GH-739]
* cli: Output of agent-info is sorted [GH-617]
* cli: Eval monitor detects zero wait condition [GH-776]
* cli: Ability to navigate allocation directories [GH-709]
* client: Handle updates to tasks Restart Policy and KillTimeout [GH-751]
* client: Create a tmp/ directory inside each task directory [GH-757]
@@ -34,6 +36,8 @@ IMPROVEMENTS:
* drivers: Interpret Nomad variables in environment variables/args [GH-653]
* driver/rkt: Add support for CPU/Memory isolation [GH-610]
* driver/rkt: Add support for mounting alloc/task directory [GH-645]
* driver/docker: Support for .dockercfg based auth for private registries
[GH-773]
BUG FIXES:
* core: Node drain could only be partially applied [GH-750]
@@ -42,6 +46,7 @@ BUG FIXES:
* client: Handle non-200 codes when parsing AWS metadata [GH-614]
* client: Cleanup of the allocation directory [GH-755]
* client: Unmounted of shared alloc dir when client is rebooted [GH-755]
* client/consul: Service name changes handled properly [GH-766]
* driver/rkt: handle broader format of rkt version outputs [GH-745]
* driver/qemu: failed to load image and kvm accelerator fixes [GH-656]

2
README.md
View File

@@ -57,7 +57,7 @@ Developing Nomad
If you wish to work on Nomad itself or any of its built-in systems,
you will first need [Go](https://www.golang.org) installed on your
machine (version 1.4+ is *required*).
machine (version 1.5+ is *required*).
**Developing with Vagrant**
There is an included Vagrantfile that can help bootstrap the process. The

2
Vagrantfile vendored
View File

@@ -62,6 +62,8 @@ sudo usermod -aG docker vagrant
# Setup Nomad for development
cd /opt/gopath/src/github.com/hashicorp/nomad && make updatedeps
# Install gox
go get github.com/mitchellh/gox
# CD into the nomad working directory when we login to the VM
grep "cd /opt/gopath/src/github.com/hashicorp/nomad" ~/.profile || echo "cd /opt/gopath/src/github.com/hashicorp/nomad" >> ~/.profile

18
client/alloc_runner.go
View File

@@ -364,8 +364,6 @@ func (r *AllocRunner) Run() {
continue
}
// Merge in the task resources
task.Resources = alloc.TaskResources[task.Name]
tr := NewTaskRunner(r.logger, r.config, r.setTaskState, r.ctx, r.alloc,
task, r.consulService)
r.tasks[task.Name] = tr
@@ -392,22 +390,6 @@ OUTER:
r.taskLock.RLock()
for _, task := range tg.Tasks {
tr := r.tasks[task.Name]
// Merge in the task resources
task.Resources = update.TaskResources[task.Name]
FOUND:
for _, updateGroup := range update.Job.TaskGroups {
if tg.Name != updateGroup.Name {
continue
}
for _, updateTask := range updateGroup.Tasks {
if updateTask.Name != task.Name {
continue
}
task.Services = updateTask.Services
break FOUND
}
}
tr.Update(update)
}
r.taskLock.RUnlock()

6
client/config/config.go
View File

@@ -57,6 +57,12 @@ type Config struct {
// Node provides the base node
Node *structs.Node
// ExecutorMaxPort defines the highest port a plugin process can use
ExecutorMaxPort int
// ExecutorMinPort defines the lowest port a plugin process can use
ExecutorMinPort int
// Options provides arbitrary key-value configuration for nomad internals,
// like fingerprinters and drivers. The format is:
//

31
client/consul.go
View File

@@ -76,6 +76,7 @@ type ConsulService struct {
trackedTasks map[string]*trackedTask
serviceStates map[string]string
allocToService map[string][]string
trackedTskLock sync.Mutex
}
@@ -130,12 +131,13 @@ func NewConsulService(config *consulServiceConfig) (*ConsulService, error) {
}
consulService := ConsulService{
client: &consulApiClient{client: c},
logger: config.logger,
node: config.node,
trackedTasks: make(map[string]*trackedTask),
serviceStates: make(map[string]string),
shutdownCh: make(chan struct{}),
client: &consulApiClient{client: c},
logger: config.logger,
node: config.node,
trackedTasks: make(map[string]*trackedTask),
serviceStates: make(map[string]string),
allocToService: make(map[string][]string),
shutdownCh: make(chan struct{}),
}
return &consulService, nil
@@ -148,8 +150,18 @@ func (c *ConsulService) Register(task *structs.Task, alloc *structs.Allocation)
c.trackedTskLock.Lock()
tt := &trackedTask{task: task, alloc: alloc}
c.trackedTasks[fmt.Sprintf("%s-%s", alloc.ID, task.Name)] = tt
// Delete any previously registered service as the same alloc is being
// re-registered.
for _, service := range c.allocToService[alloc.ID] {
delete(c.serviceStates, service)
}
c.trackedTskLock.Unlock()
for _, service := range task.Services {
// Track the services this alloc is registering.
c.allocToService[alloc.ID] = append(c.allocToService[alloc.ID], service.Name)
c.logger.Printf("[INFO] consul: registering service %s with consul.", service.Name)
if err := c.registerService(service, task, alloc); err != nil {
mErr.Errors = append(mErr.Errors, err)
@@ -165,6 +177,7 @@ func (c *ConsulService) Deregister(task *structs.Task, alloc *structs.Allocation
var mErr multierror.Error
c.trackedTskLock.Lock()
delete(c.trackedTasks, fmt.Sprintf("%s-%s", alloc.ID, task.Name))
delete(c.allocToService, alloc.ID)
c.trackedTskLock.Unlock()
for _, service := range task.Services {
serviceID := alloc.Services[service.Name]
@@ -229,14 +242,14 @@ func (c *ConsulService) performSync() {
// Add new services which Consul agent isn't aware of
knownServices[serviceID] = struct{}{}
if _, ok := consulServices[serviceID]; !ok {
c.printLogMessage("[INFO] consul: registering service %s with consul.", service.Name)
c.printLogMessage("[INFO] consul: perform sync, registering service %s with consul.", service.Name)
c.registerService(service, trackedTask.task, trackedTask.alloc)
continue
}
// If a service has changed, re-register it with Consul agent
if service.Hash() != c.serviceStates[serviceID] {
c.printLogMessage("[INFO] consul: reregistering service %s with consul.", service.Name)
c.printLogMessage("[INFO] consul: perform sync hash change, reregistering service %s with consul.", service.Name)
c.registerService(service, trackedTask.task, trackedTask.alloc)
continue
}
@@ -268,7 +281,7 @@ func (c *ConsulService) performSync() {
for _, consulService := range consulServices {
if _, ok := knownServices[consulService.ID]; !ok {
delete(c.serviceStates, consulService.ID)
c.printLogMessage("[INFO] consul: deregistering service %v with consul", consulService.Service)
c.printLogMessage("[INFO] consul: perform sync, deregistering service %v with consul", consulService.Service)
c.deregisterService(consulService.ID)
}
}

12
client/driver/docker.go
View File

@@ -5,6 +5,7 @@ import (
"fmt"
"log"
"net"
"os"
"path/filepath"
"strconv"
"strings"
@@ -448,6 +449,17 @@ func (d *DockerDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle
}
}
if authConfig := d.config.Read("docker.auth.config"); authConfig != "" {
if f, err := os.Open(authConfig); err == nil {
defer f.Close()
if authConfigurations, err := docker.NewAuthConfigurations(f); err == nil {
if authConfiguration, ok := authConfigurations.Configs[repo]; ok {
authOptions = authConfiguration
}
}
}
}
err = client.PullImage(pullOptions, authOptions)
if err != nil {
d.logger.Printf("[ERR] driver.docker: failed pulling container %s:%s: %s", repo, tag, err)

21
client/driver/exec.go
View File

@@ -17,6 +17,8 @@ import (
"github.com/hashicorp/nomad/helper/discover"
"github.com/hashicorp/nomad/nomad/structs"
"github.com/mitchellh/mapstructure"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
)
// ExecDriver fork/execs tasks using as many of the underlying OS's isolation
@@ -36,6 +38,7 @@ type ExecDriverConfig struct {
type execHandle struct {
pluginClient *plugin.Client
executor executor.Executor
groups *cgroupConfig.Cgroup
userPid int
killTimeout time.Duration
logger *log.Logger
@@ -102,7 +105,7 @@ func (d *ExecDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
if err != nil {
return nil, fmt.Errorf("unable to find the nomad binary: %v", err)
}
pluginLogFile := filepath.Join(ctx.AllocDir.AllocDir, "plugin.out")
pluginLogFile := filepath.Join(taskDir, fmt.Sprintf("%s-executor.out", task.Name))
pluginConfig := &plugin.ClientConfig{
Cmd: exec.Command(bin, "executor", pluginLogFile),
}
@@ -126,13 +129,14 @@ func (d *ExecDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
pluginClient.Kill()
return nil, fmt.Errorf("error starting process via the plugin: %v", err)
}
d.logger.Printf("started process via plugin with pid: %v", ps.Pid)
d.logger.Printf("[DEBUG] driver.exec: started process via plugin with pid: %v", ps.Pid)
// Return a driver handle
h := &execHandle{
pluginClient: pluginClient,
userPid: ps.Pid,
executor: exec,
groups: &ps.IsolationConfig,
killTimeout: d.DriverContext.KillTimeout(task),
logger: d.logger,
doneCh: make(chan struct{}),
@@ -145,6 +149,7 @@ func (d *ExecDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
type execId struct {
KillTimeout time.Duration
UserPid int
Groups *cgroupConfig.Cgroup
PluginConfig *ExecutorReattachConfig
}
@@ -154,15 +159,17 @@ func (d *ExecDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, erro
return nil, fmt.Errorf("Failed to parse handle '%s': %v", handleID, err)
}
reattachConfig := id.PluginConfig.PluginConfig()
pluginConfig := &plugin.ClientConfig{
Reattach: reattachConfig,
Reattach: id.PluginConfig.PluginConfig(),
}
executor, client, err := createExecutor(pluginConfig, d.config.LogOutput)
if err != nil {
d.logger.Println("[ERROR] error connecting to plugin so destroying plugin pid and user pid")
d.logger.Println("[ERROR] driver.exec: error connecting to plugin so destroying plugin pid and user pid")
if e := destroyPlugin(id.PluginConfig.Pid, id.UserPid); e != nil {
d.logger.Printf("[ERROR] error destroying plugin and userpid: %v", e)
d.logger.Printf("[ERROR] driver.exec: error destroying plugin and userpid: %v", e)
}
if e := destroyCgroup(id.Groups); e != nil {
d.logger.Printf("[ERROR] driver.exec: %v", e)
}
return nil, fmt.Errorf("error connecting to plugin: %v", err)
}
@@ -172,6 +179,7 @@ func (d *ExecDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, erro
pluginClient: client,
executor: executor,
userPid: id.UserPid,
groups: id.Groups,
logger: d.logger,
killTimeout: id.KillTimeout,
doneCh: make(chan struct{}),
@@ -186,6 +194,7 @@ func (h *execHandle) ID() string {
KillTimeout: h.killTimeout,
PluginConfig: NewExecutorReattachConfig(h.pluginClient.ReattachConfig()),
UserPid: h.userPid,
Groups: h.groups,
}
data, err := json.Marshal(id)

59
client/driver/exec_test.go
View File

@@ -1,10 +1,13 @@
package driver
import (
"encoding/json"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"reflect"
"syscall"
"testing"
"time"
@@ -75,6 +78,62 @@ func TestExecDriver_StartOpen_Wait(t *testing.T) {
handle2.Kill()
}
func TestExecDriver_KillUserPid_OnPluginReconnectFailure(t *testing.T) {
t.Parallel()
ctestutils.ExecCompatible(t)
task := &structs.Task{
Name: "sleep",
Config: map[string]interface{}{
"command": "/bin/sleep",
"args": []string{"1000000"},
},
Resources: basicResources,
}
driverCtx, execCtx := testDriverContexts(task)
defer execCtx.AllocDir.Destroy()
d := NewExecDriver(driverCtx)
handle, err := d.Start(execCtx, task)
defer handle.Kill()
if err != nil {
t.Fatalf("err: %v", err)
}
if handle == nil {
t.Fatalf("missing handle")
}
id := &execId{}
if err := json.Unmarshal([]byte(handle.ID()), id); err != nil {
t.Fatalf("Failed to parse handle '%s': %v", handle.ID(), err)
}
pluginPid := id.PluginConfig.Pid
proc, err := os.FindProcess(pluginPid)
if err != nil {
t.Fatalf("can't find plugin pid: %v", pluginPid)
}
if err := proc.Kill(); err != nil {
t.Fatalf("can't kill plugin pid: %v", err)
}
// Attempt to open
handle2, err := d.Open(execCtx, handle.ID())
if err == nil {
t.Fatalf("expected error")
}
if handle2 != nil {
handle2.Kill()
t.Fatalf("expected handle2 to be nil")
}
// Test if the userpid is still present
userProc, err := os.FindProcess(id.UserPid)
err = userProc.Signal(syscall.Signal(0))
if err != nil {
t.Fatalf("expected user process to die")
}
}
func TestExecDriver_Start_Wait(t *testing.T) {
t.Parallel()
ctestutils.ExecCompatible(t)

107
client/driver/executor/executor.go
View File

@@ -8,10 +8,12 @@ import (
"os/exec"
"path/filepath"
"runtime"
"strings"
"sync"
"syscall"
"time"
"github.com/hashicorp/go-multierror"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
"github.com/hashicorp/nomad/client/allocdir"
@@ -20,35 +22,56 @@ import (
"github.com/hashicorp/nomad/nomad/structs"
)
// ExecutorContext is a wrapper to hold context to configure the command user
// wants to run
// ExecutorContext holds context to configure the command user
// wants to run and isolate it
type ExecutorContext struct {
TaskEnv *env.TaskEnvironment
AllocDir *allocdir.AllocDir
TaskName string
TaskResources *structs.Resources
LogConfig *structs.LogConfig
FSIsolation bool
ResourceLimits bool
// TaskEnv holds information about the environment of a Task
TaskEnv *env.TaskEnvironment
// AllocDir is the handle to do operations on the alloc dir of
// the task
AllocDir *allocdir.AllocDir
// TaskName is the name of the Task
TaskName string
// TaskResources are the resource constraints for the Task
TaskResources *structs.Resources
// FSIsolation is a flag for drivers to impose file system
// isolation on certain platforms
FSIsolation bool
// ResourceLimits is a flag for drivers to impose resource
// contraints on a Task on certain platforms
ResourceLimits bool
// UnprivilegedUser is a flag for drivers to make the process
// run as nobody
UnprivilegedUser bool
// LogConfig provides the configuration related to log rotation
LogConfig *structs.LogConfig
}
// ExecCommand is a wrapper to hold the user command
// ExecCommand holds the user command and args. It's a lightweight replacement
// of exec.Cmd for serialization purposes.
type ExecCommand struct {
Cmd string
Args []string
}
// ProcessState holds information about the state of
// a user process
// ProcessState holds information about the state of a user process.
type ProcessState struct {
Pid int
ExitCode int
Time time.Time
Pid int
ExitCode int
Signal int
IsolationConfig cgroupConfig.Cgroup
Time time.Time
}
// Executor is the interface which allows a driver to launch and supervise
// a process user wants to run
// a process
type Executor interface {
LaunchCmd(command *ExecCommand, ctx *ExecutorContext) (*ProcessState, error)
Wait() (*ProcessState, error)
@@ -83,17 +106,22 @@ func NewExecutor(logger *log.Logger) Executor {
// LaunchCmd launches a process and returns it's state. It also configures an
// applies isolation on certain platforms.
func (e *UniversalExecutor) LaunchCmd(command *ExecCommand, ctx *ExecutorContext) (*ProcessState, error) {
e.logger.Printf("[INFO] executor: launching command %v", command.Cmd)
e.logger.Printf("[DEBUG] executor: launching command %v %v", command.Cmd, strings.Join(command.Args, ""))
e.ctx = ctx
// configuring the task dir
if err := e.configureTaskDir(); err != nil {
return nil, err
}
// configuring the chroot, cgroup and enters the plugin process in the
// chroot
if err := e.configureIsolation(); err != nil {
return nil, err
}
// setting the user of the process
if e.ctx.UnprivilegedUser {
if err := e.runAs("nobody"); err != nil {
return nil, err
@@ -103,7 +131,9 @@ func (e *UniversalExecutor) LaunchCmd(command *ExecCommand, ctx *ExecutorContext
logFileSize := int64(ctx.LogConfig.MaxFileSizeMB * 1024 * 1024)
stdor, stdow := io.Pipe()
lro, err := logrotator.NewLogRotator(filepath.Join(e.taskDir, allocdir.TaskLocal), fmt.Sprintf("%v.stdout", ctx.TaskName), ctx.LogConfig.MaxFiles, logFileSize, e.logger)
lro, err := logrotator.NewLogRotator(filepath.Join(e.taskDir, allocdir.TaskLocal),
fmt.Sprintf("%v.stdout", ctx.TaskName), ctx.LogConfig.MaxFiles,
logFileSize, e.logger)
if err != nil {
return nil, fmt.Errorf("error creating log rotator for stdout of task %v", err)
}
@@ -112,7 +142,9 @@ func (e *UniversalExecutor) LaunchCmd(command *ExecCommand, ctx *ExecutorContext
go lro.Start(stdor)
stder, stdew := io.Pipe()
lre, err := logrotator.NewLogRotator(filepath.Join(e.taskDir, allocdir.TaskLocal), fmt.Sprintf("%v.stderr", ctx.TaskName), ctx.LogConfig.MaxFiles, logFileSize, e.logger)
lre, err := logrotator.NewLogRotator(filepath.Join(e.taskDir, allocdir.TaskLocal),
fmt.Sprintf("%v.stderr", ctx.TaskName), ctx.LogConfig.MaxFiles,
logFileSize, e.logger)
if err != nil {
return nil, fmt.Errorf("error creating log rotator for stderr of task %v", err)
}
@@ -120,8 +152,9 @@ func (e *UniversalExecutor) LaunchCmd(command *ExecCommand, ctx *ExecutorContext
e.lre = lre
go lre.Start(stder)
// setting the env, path and args for the command
e.ctx.TaskEnv.Build()
e.cmd.Env = ctx.TaskEnv.EnvList()
e.cmd.Path = ctx.TaskEnv.ReplaceEnv(command.Cmd)
e.cmd.Args = append([]string{e.cmd.Path}, ctx.TaskEnv.ParseAndReplace(command.Args)...)
if filepath.Base(command.Cmd) == command.Cmd {
@@ -131,13 +164,13 @@ func (e *UniversalExecutor) LaunchCmd(command *ExecCommand, ctx *ExecutorContext
}
}
// starting the process
if err := e.cmd.Start(); err != nil {
return nil, fmt.Errorf("error starting command: %v", err)
}
e.applyLimits()
go e.wait()
return &ProcessState{Pid: e.cmd.Process.Pid, ExitCode: -1, Time: time.Now()}, nil
return &ProcessState{Pid: e.cmd.Process.Pid, ExitCode: -1, IsolationConfig: *e.groups, Time: time.Now()}, nil
}
// Wait waits until a process has exited and returns it's exitcode and errors
@@ -187,24 +220,28 @@ func (e *UniversalExecutor) wait() {
// Exit cleans up the alloc directory, destroys cgroups and kills the user
// process
func (e *UniversalExecutor) Exit() error {
e.logger.Printf("[INFO] Exiting plugin for task %q", e.ctx.TaskName)
if e.cmd.Process == nil {
return fmt.Errorf("executor.exit error: no process found")
}
proc, err := os.FindProcess(e.cmd.Process.Pid)
if err != nil {
return fmt.Errorf("failied to find user process %v: %v", e.cmd.Process.Pid, err)
var merr multierror.Error
if e.cmd.Process != nil {
proc, err := os.FindProcess(e.cmd.Process.Pid)
if err != nil {
e.logger.Printf("[ERROR] can't find process with pid: %v, err: %v", e.cmd.Process.Pid, err)
}
if err := proc.Kill(); err != nil {
e.logger.Printf("[ERROR] can't kill process with pid: %v, err: %v", e.cmd.Process.Pid, err)
}
}
if e.ctx.FSIsolation {
e.removeChrootMounts()
if err := e.removeChrootMounts(); err != nil {
merr.Errors = append(merr.Errors, err)
}
}
if e.ctx.ResourceLimits {
e.destroyCgroup()
if err := e.destroyCgroup(); err != nil {
merr.Errors = append(merr.Errors, err)
}
}
if err = proc.Kill(); err != nil {
e.logger.Printf("[DEBUG] executor.exit error: %v", err)
}
return nil
return merr.ErrorOrNil()
}
// Shutdown sends an interrupt signal to the user process

2
client/driver/executor/executor_basic.go
View File

@@ -18,7 +18,7 @@ func (e *UniversalExecutor) runAs(userid string) error {
return nil
}
func (e *UniversalExecutor) applyLimits() error {
func (e *UniversalExecutor) applyLimits(pid int) error {
return nil
}

19
client/driver/executor/executor_linux.go
View File

@@ -45,23 +45,35 @@ func (e *UniversalExecutor) configureIsolation() error {
if err := e.configureCgroups(e.ctx.TaskResources); err != nil {
return fmt.Errorf("error creating cgroups: %v", err)
}
if err := e.applyLimits(os.Getpid()); err != nil {
if er := e.destroyCgroup(); er != nil {
e.logger.Printf("[ERROR] error destroying cgroup: %v", er)
}
if er := e.removeChrootMounts(); er != nil {
e.logger.Printf("[ERROR] error removing chroot: %v", er)
}
return fmt.Errorf("error entering the plugin process in the cgroup: %v:", err)
}
}
return nil
}
// applyLimits puts a process in a pre-configured cgroup
func (e *UniversalExecutor) applyLimits() error {
func (e *UniversalExecutor) applyLimits(pid int) error {
if !e.ctx.ResourceLimits {
return nil
}
// Entering the process in the cgroup
manager := e.getCgroupManager(e.groups)
if err := manager.Apply(e.cmd.Process.Pid); err != nil {
if err := manager.Apply(pid); err != nil {
e.logger.Printf("[ERROR] unable to join cgroup: %v", err)
if err := e.Exit(); err != nil {
e.logger.Printf("[ERROR] unable to kill process: %v", err)
}
return err
}
return nil
}
@@ -210,8 +222,7 @@ func (e *UniversalExecutor) removeChrootMounts() error {
}
}
// Unmount
// proc.
// Unmount proc.
proc := filepath.Join(e.taskDir, "proc")
if e.pathExists(proc) {
if err := syscall.Unmount(proc, 0); err != nil {

55
client/driver/executor/executor_test.go
View File

@@ -11,9 +11,10 @@ import (
"github.com/hashicorp/nomad/client/allocdir"
"github.com/hashicorp/nomad/client/driver/env"
"github.com/hashicorp/nomad/client/testutil"
"github.com/hashicorp/nomad/nomad/mock"
"github.com/hashicorp/nomad/nomad/structs"
"github.com/hashicorp/nomad/testutil"
tu "github.com/hashicorp/nomad/testutil"
)
var (
@@ -57,17 +58,18 @@ func TestExecutor_Start_Invalid(t *testing.T) {
invalid := "/bin/foobar"
execCmd := ExecCommand{Cmd: invalid, Args: []string{"1"}}
ctx := testExecutorContext(t)
defer ctx.AllocDir.Destroy()
executor := NewExecutor(log.New(os.Stdout, "", log.LstdFlags))
_, err := executor.LaunchCmd(&execCmd, ctx)
if err == nil {
t.Fatalf("Expected error")
}
defer ctx.AllocDir.Destroy()
}
func TestExecutor_Start_Wait_Failure_Code(t *testing.T) {
execCmd := ExecCommand{Cmd: "/bin/sleep", Args: []string{"fail"}}
ctx := testExecutorContext(t)
defer ctx.AllocDir.Destroy()
executor := NewExecutor(log.New(os.Stdout, "", log.LstdFlags))
ps, _ := executor.LaunchCmd(&execCmd, ctx)
if ps.Pid == 0 {
@@ -77,12 +79,12 @@ func TestExecutor_Start_Wait_Failure_Code(t *testing.T) {
if ps.ExitCode < 1 {
t.Fatalf("expected exit code to be non zero, actual: %v", ps.ExitCode)
}
defer ctx.AllocDir.Destroy()
}
func TestExecutor_Start_Wait(t *testing.T) {
execCmd := ExecCommand{Cmd: "/bin/echo", Args: []string{"hello world"}}
ctx := testExecutorContext(t)
defer ctx.AllocDir.Destroy()
executor := NewExecutor(log.New(os.Stdout, "", log.LstdFlags))
ps, err := executor.LaunchCmd(&execCmd, ctx)
if err != nil {
@@ -95,8 +97,51 @@ func TestExecutor_Start_Wait(t *testing.T) {
if err != nil {
t.Fatalf("error in waiting for command: %v", err)
}
task := "web"
taskDir, ok := ctx.AllocDir.TaskDirs[task]
if !ok {
log.Panicf("No task directory found for task %v", task)
}
expected := "hello world"
file := filepath.Join(allocdir.TaskLocal, "web.stdout")
absFilePath := filepath.Join(taskDir, file)
output, err := ioutil.ReadFile(absFilePath)
if err != nil {
t.Fatalf("Couldn't read file %v", absFilePath)
}
act := strings.TrimSpace(string(output))
if act != expected {
t.Fatalf("Command output incorrectly: want %v; got %v", expected, act)
}
}
func TestExecutor_IsolationAndConstraints(t *testing.T) {
testutil.ExecCompatible(t)
execCmd := ExecCommand{Cmd: "/bin/echo", Args: []string{"hello world"}}
ctx := testExecutorContext(t)
defer ctx.AllocDir.Destroy()
ctx.FSIsolation = true
ctx.ResourceLimits = true
ctx.UnprivilegedUser = true
executor := NewExecutor(log.New(os.Stdout, "", log.LstdFlags))
ps, err := executor.LaunchCmd(&execCmd, ctx)
if err != nil {
t.Fatalf("error in launching command: %v", err)
}
if ps.Pid == 0 {
t.Fatalf("expected process to start and have non zero pid")
}
ps, err = executor.Wait()
if err != nil {
t.Fatalf("error in waiting for command: %v", err)
}
task := "web"
taskDir, ok := ctx.AllocDir.TaskDirs[task]
if !ok {
@@ -120,6 +165,7 @@ func TestExecutor_Start_Wait(t *testing.T) {
func TestExecutor_Start_Kill(t *testing.T) {
execCmd := ExecCommand{Cmd: "/bin/sleep", Args: []string{"10 && hello world"}}
ctx := testExecutorContext(t)
defer ctx.AllocDir.Destroy()
executor := NewExecutor(log.New(os.Stdout, "", log.LstdFlags))
ps, err := executor.LaunchCmd(&execCmd, ctx)
if err != nil {
@@ -132,7 +178,6 @@ func TestExecutor_Start_Kill(t *testing.T) {
if err != nil {
t.Fatalf("error in waiting for command: %v", err)
}
defer ctx.AllocDir.Destroy()
task := "web"
taskDir, ok := ctx.AllocDir.TaskDirs[task]
@@ -143,7 +188,7 @@ func TestExecutor_Start_Kill(t *testing.T) {
file := filepath.Join(allocdir.TaskLocal, "web.stdout")
absFilePath := filepath.Join(taskDir, file)
time.Sleep(time.Duration(testutil.TestMultiplier()*2) * time.Second)
time.Sleep(time.Duration(tu.TestMultiplier()*2) * time.Second)
output, err := ioutil.ReadFile(absFilePath)
if err != nil {

32
client/driver/java.go
View File

@@ -14,6 +14,7 @@ import (
"github.com/hashicorp/go-plugin"
"github.com/mitchellh/mapstructure"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
"github.com/hashicorp/nomad/client/config"
"github.com/hashicorp/nomad/client/driver/executor"
@@ -43,6 +44,7 @@ type javaHandle struct {
pluginClient *plugin.Client
userPid int
executor executor.Executor
groups *cgroupConfig.Cgroup
killTimeout time.Duration
logger *log.Logger
@@ -148,7 +150,7 @@ func (d *JavaDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
return nil, fmt.Errorf("unable to find the nomad binary: %v", err)
}
pluginLogFile := filepath.Join(ctx.AllocDir.AllocDir, "plugin.out")
pluginLogFile := filepath.Join(taskDir, fmt.Sprintf("%s-executor.out", task.Name))
pluginConfig := &plugin.ClientConfig{
Cmd: exec.Command(bin, "executor", pluginLogFile),
}
@@ -158,24 +160,26 @@ func (d *JavaDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
return nil, err
}
executorCtx := &executor.ExecutorContext{
TaskEnv: d.taskEnv,
AllocDir: ctx.AllocDir,
TaskName: task.Name,
TaskResources: task.Resources,
LogConfig: task.LogConfig,
TaskEnv: d.taskEnv,
AllocDir: ctx.AllocDir,
TaskName: task.Name,
TaskResources: task.Resources,
UnprivilegedUser: true,
LogConfig: task.LogConfig,
}
ps, err := exec.LaunchCmd(&executor.ExecCommand{Cmd: "java", Args: args}, executorCtx)
if err != nil {
pluginClient.Kill()
return nil, fmt.Errorf("error starting process via the plugin: %v", err)
}
d.logger.Printf("[INFO] started process with pid: %v", ps.Pid)
d.logger.Printf("[DEBUG] driver.java: started process with pid: %v", ps.Pid)
// Return a driver handle
h := &javaHandle{
pluginClient: pluginClient,
executor: exec,
userPid: ps.Pid,
groups: &ps.IsolationConfig,
killTimeout: d.DriverContext.KillTimeout(task),
logger: d.logger,
doneCh: make(chan struct{}),
@@ -189,6 +193,7 @@ func (d *JavaDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
type javaId struct {
KillTimeout time.Duration
PluginConfig *ExecutorReattachConfig
Groups *cgroupConfig.Cgroup
UserPid int
}
@@ -198,16 +203,19 @@ func (d *JavaDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, erro
return nil, fmt.Errorf("Failed to parse handle '%s': %v", handleID, err)
}
reattachConfig := id.PluginConfig.PluginConfig()
pluginConfig := &plugin.ClientConfig{
Reattach: reattachConfig,
Reattach: id.PluginConfig.PluginConfig(),
}
executor, pluginClient, err := createExecutor(pluginConfig, d.config.LogOutput)
if err != nil {
d.logger.Println("[ERROR] error connecting to plugin so destroying plugin pid and user pid")
d.logger.Println("[ERROR] driver.java: error connecting to plugin so destroying plugin pid and user pid")
if e := destroyPlugin(id.PluginConfig.Pid, id.UserPid); e != nil {
d.logger.Printf("[ERROR] error destroying plugin and userpid: %v", e)
d.logger.Printf("[ERROR] driver.java: error destroying plugin and userpid: %v", e)
}
if e := destroyCgroup(id.Groups); e != nil {
d.logger.Printf("[ERROR] driver.exec: %v", e)
}
return nil, fmt.Errorf("error connecting to plugin: %v", err)
}
@@ -216,6 +224,7 @@ func (d *JavaDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, erro
pluginClient: pluginClient,
executor: executor,
userPid: id.UserPid,
groups: id.Groups,
logger: d.logger,
killTimeout: id.KillTimeout,
doneCh: make(chan struct{}),
@@ -231,6 +240,7 @@ func (h *javaHandle) ID() string {
KillTimeout: h.killTimeout,
PluginConfig: NewExecutorReattachConfig(h.pluginClient.ReattachConfig()),
UserPid: h.userPid,
Groups: h.groups,
}
data, err := json.Marshal(id)

10
client/driver/qemu.go
View File

@@ -192,7 +192,7 @@ func (d *QemuDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
return nil, fmt.Errorf("unable to find the nomad binary: %v", err)
}
pluginLogFile := filepath.Join(ctx.AllocDir.AllocDir, "plugin.out")
pluginLogFile := filepath.Join(taskDir, fmt.Sprintf("%s-executor.out", task.Name))
pluginConfig := &plugin.ClientConfig{
Cmd: exec.Command(bin, "executor", pluginLogFile),
}
@@ -213,7 +213,6 @@ func (d *QemuDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandle,
pluginClient.Kill()
return nil, fmt.Errorf("error starting process via the plugin: %v", err)
}
d.logger.Printf("[INFO] started process with pid: %v", ps.Pid)
d.logger.Printf("[INFO] Started new QemuVM: %s", vmID)
// Create and Return Handle
@@ -243,16 +242,15 @@ func (d *QemuDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, erro
return nil, fmt.Errorf("Failed to parse handle '%s': %v", handleID, err)
}
reattachConfig := id.PluginConfig.PluginConfig()
pluginConfig := &plugin.ClientConfig{
Reattach: reattachConfig,
Reattach: id.PluginConfig.PluginConfig(),
}
executor, pluginClient, err := createExecutor(pluginConfig, d.config.LogOutput)
if err != nil {
d.logger.Println("[ERROR] error connecting to plugin so destroying plugin pid and user pid")
d.logger.Println("[ERROR] driver.qemu: error connecting to plugin so destroying plugin pid and user pid")
if e := destroyPlugin(id.PluginConfig.Pid, id.UserPid); e != nil {
d.logger.Printf("[ERROR] error destroying plugin and userpid: %v", e)
d.logger.Printf("[ERROR] driver.qemu: error destroying plugin and userpid: %v", e)
}
return nil, fmt.Errorf("error connecting to plugin: %v", err)
}

8
client/driver/raw_exec.go
View File

@@ -98,7 +98,7 @@ func (d *RawExecDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandl
if err != nil {
return nil, fmt.Errorf("unable to find the nomad binary: %v", err)
}
pluginLogFile := filepath.Join(ctx.AllocDir.AllocDir, "plugin.out")
pluginLogFile := filepath.Join(taskDir, fmt.Sprintf("%s-executor.out", task.Name))
pluginConfig := &plugin.ClientConfig{
Cmd: exec.Command(bin, "executor", pluginLogFile),
}
@@ -119,7 +119,7 @@ func (d *RawExecDriver) Start(ctx *ExecContext, task *structs.Task) (DriverHandl
pluginClient.Kill()
return nil, fmt.Errorf("error starting process via the plugin: %v", err)
}
d.logger.Printf("[INFO] started process with pid: %v", ps.Pid)
d.logger.Printf("[DEBUG] driver.raw_exec: started process with pid: %v", ps.Pid)
// Return a driver handle
h := &rawExecHandle{
@@ -152,9 +152,9 @@ func (d *RawExecDriver) Open(ctx *ExecContext, handleID string) (DriverHandle, e
}
executor, pluginClient, err := createExecutor(pluginConfig, d.config.LogOutput)
if err != nil {
d.logger.Println("[ERROR] error connecting to plugin so destroying plugin pid and user pid")
d.logger.Println("[ERROR] driver.raw_exec: error connecting to plugin so destroying plugin pid and user pid")
if e := destroyPlugin(id.PluginConfig.Pid, id.UserPid); e != nil {
d.logger.Printf("[ERROR] error destroying plugin and userpid: %v", e)
d.logger.Printf("[ERROR] driver.raw_exec: error destroying plugin and userpid: %v", e)
}
return nil, fmt.Errorf("error connecting to plugin: %v", err)
}

8
client/driver/utils.go
View File

@@ -10,9 +10,14 @@ import (
"github.com/hashicorp/nomad/client/driver/executor"
)
// createExecutor launches an executor plugin and returns an instance of the
// Executor interface
func createExecutor(config *plugin.ClientConfig, w io.Writer) (executor.Executor, *plugin.Client, error) {
config.HandshakeConfig = HandshakeConfig
config.Plugins = GetPluginMap(w)
if config.Cmd != nil {
isolateCommand(config.Cmd)
}
executorClient := plugin.NewClient(config)
rpcClient, err := executorClient.Client()
if err != nil {
@@ -27,6 +32,7 @@ func createExecutor(config *plugin.ClientConfig, w io.Writer) (executor.Executor
return executorPlugin, executorClient, nil
}
// killProcess kills a process with the given pid
func killProcess(pid int) error {
proc, err := os.FindProcess(pid)
if err != nil {
@@ -35,6 +41,8 @@ func killProcess(pid int) error {
return proc.Kill()
}
// destroyPlugin kills the plugin with the given pid and also kills the user
// process
func destroyPlugin(pluginPid int, userPid int) error {
var merr error
if err := killProcess(pluginPid); err != nil {

40
client/driver/utils_linux.go Normal file
View File

@@ -0,0 +1,40 @@
package driver
import (
"os/exec"
"syscall"
"fmt"
"github.com/opencontainers/runc/libcontainer/cgroups"
cgroupFs "github.com/opencontainers/runc/libcontainer/cgroups/fs"
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
)
// isolateCommand sets the setsid flag in exec.Cmd to true so that the process
// becomes the process leader in a new session and doesn't receive signals that
// are sent to the parent process.
func isolateCommand(cmd *exec.Cmd) {
if cmd.SysProcAttr == nil {
cmd.SysProcAttr = &syscall.SysProcAttr{}
}
cmd.SysProcAttr.Setsid = true
}
// destroyCgroup destroys a cgroup and thereby killing all the processes in that
// group
func destroyCgroup(group *cgroupConfig.Cgroup) error {
if group == nil {
return nil
}
var manager cgroups.Manager
manager = &cgroupFs.Manager{Cgroups: group}
if systemd.UseSystemd() {
manager = &systemd.Manager{Cgroups: group}
}
if err := manager.Destroy(); err != nil {
return fmt.Errorf("failed to destroy cgroup: %v", err)
}
return nil
}

24
client/driver/utils_posix.go Normal file
View File

@@ -0,0 +1,24 @@
// +build !linux
package driver
import (
"os/exec"
"syscall"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
)
// isolateCommand sets the setsid flag in exec.Cmd to true so that the process
// becomes the process leader in a new session and doesn't receive signals that
// are sent to the parent process.
func isolateCommand(cmd *exec.Cmd) {
if cmd.SysProcAttr == nil {
cmd.SysProcAttr = &syscall.SysProcAttr{}
}
cmd.SysProcAttr.Setsid = true
}
func destroyCgroup(group *cgroupConfig.Cgroup) error {
return nil
}

15
client/driver/utils_windows.go Normal file
View File

@@ -0,0 +1,15 @@
package driver
import (
"os/exec"
cgroupConfig "github.com/opencontainers/runc/libcontainer/configs"
)
// TODO Figure out if this is needed in Wondows
func isolateCommand(cmd *exec.Cmd) {
}
func destroyCgroup(group *cgroupConfig.Cgroup) error {
return nil
}

46
client/task_runner.go
View File

@@ -10,9 +10,11 @@ import (
"sync"
"time"
"github.com/hashicorp/go-multierror"
"github.com/hashicorp/nomad/client/config"
"github.com/hashicorp/nomad/client/driver"
"github.com/hashicorp/nomad/nomad/structs"
"github.com/mitchellh/hashstructure"
cstructs "github.com/hashicorp/nomad/client/driver/structs"
)
@@ -54,6 +56,9 @@ func NewTaskRunner(logger *log.Logger, config *config.Config,
alloc *structs.Allocation, task *structs.Task,
consulService *ConsulService) *TaskRunner {
// Merge in the task resources
task.Resources = alloc.TaskResources[task.Name]
// Build the restart tracker.
tg := alloc.Job.LookupTaskGroup(alloc.TaskGroup)
if tg == nil {
@@ -319,21 +324,24 @@ func (r *TaskRunner) handleUpdate(update *structs.Allocation) error {
}
// Extract the task.
var task *structs.Task
var updatedTask *structs.Task
for _, t := range tg.Tasks {
if t.Name == r.task.Name {
task = t
updatedTask = t
}
}
if task == nil {
if updatedTask == nil {
return fmt.Errorf("task group %q doesn't contain task %q", tg.Name, r.task.Name)
}
r.task = task
// Merge in the task resources
updatedTask.Resources = update.TaskResources[updatedTask.Name]
// Update will update resources and store the new kill timeout.
var mErr multierror.Error
if r.handle != nil {
if err := r.handle.Update(task); err != nil {
r.logger.Printf("[ERR] client: failed to update task '%s' for alloc '%s': %v", r.task.Name, r.alloc.ID, err)
if err := r.handle.Update(updatedTask); err != nil {
mErr.Errors = append(mErr.Errors, fmt.Errorf("updating task resources failed: %v", err))
}
}
@@ -342,14 +350,26 @@ func (r *TaskRunner) handleUpdate(update *structs.Allocation) error {
r.restartTracker.SetPolicy(tg.RestartPolicy)
}
/* TODO
// Re-register the task to consul and store the updated alloc.
r.consulService.Deregister(r.task, r.alloc)
r.alloc = update
r.consulService.Register(r.task, r.alloc)
*/
// Hash services returns the hash of the task's services
hashServices := func(task *structs.Task) uint64 {
h, err := hashstructure.Hash(task.Services, nil)
if err != nil {
mErr.Errors = append(mErr.Errors, fmt.Errorf("hashing services failed %#v: %v", task.Services, err))
}
return h
}
return nil
// Re-register the task to consul if any of the services have changed.
if hashServices(updatedTask) != hashServices(r.task) {
if err := r.consulService.Register(updatedTask, update); err != nil {
mErr.Errors = append(mErr.Errors, fmt.Errorf("updating services with consul failed: %v", err))
}
}
// Store the updated alloc.
r.alloc = update
r.task = updatedTask
return mErr.ErrorOrNil()
}
// Helper function for converting a WaitResult into a TaskTerminated event.

53
command/agent/agent.go
View File

@@ -7,6 +7,7 @@ import (
"net"
"os"
"path/filepath"
"runtime"
"sync"
"time"
@@ -238,6 +239,36 @@ func (a *Agent) setupClient() error {
}
conf.Node.HTTPAddr = httpAddr
// Reserve some ports for the plugins
if runtime.GOOS == "windows" {
deviceName, err := a.findLoopbackDevice()
if conf.ExecutorMaxPort == 0 {
conf.ExecutorMaxPort = 15000
}
if conf.ExecutorMinPort == 0 {
conf.ExecutorMinPort = 14000
}
if err != nil {
return fmt.Errorf("error finding the device name for the ip 127.0.0.1: %v", err)
}
var nr *structs.NetworkResource
for _, n := range conf.Node.Reserved.Networks {
if n.Device == deviceName {
nr = n
}
}
if nr == nil {
nr = &structs.NetworkResource{
Device: deviceName,
ReservedPorts: make([]structs.Port, 0),
}
}
for i := conf.ExecutorMinPort; i <= conf.ExecutorMaxPort; i++ {
nr.ReservedPorts = append(nr.ReservedPorts, structs.Port{Label: fmt.Sprintf("plugin-%d", i), Value: i})
}
}
// Create the client
client, err := client.NewClient(conf)
if err != nil {
@@ -247,6 +278,28 @@ func (a *Agent) setupClient() error {
return nil
}
func (a *Agent) findLoopbackDevice() (string, error) {
var ifcs []net.Interface
var err error
var deviceName string
ifcs, err = net.Interfaces()
if err != nil {
return "", err
}
for _, ifc := range ifcs {
addrs, err := ifc.Addrs()
if err != nil {
return deviceName, err
}
for _, addr := range addrs {
if net.ParseIP(addr.String()).IsLoopback() {
return ifc.Name, nil
}
}
}
return deviceName, err
}
// Leave is used gracefully exit. Clients will inform servers
// of their departure so that allocations can be rescheduled.
func (a *Agent) Leave() error {

19
command/fs.go Normal file
View File

@@ -0,0 +1,19 @@
package command
import "github.com/mitchellh/cli"
type FSCommand struct {
Meta
}
func (f *FSCommand) Help() string {
return "This command is accessed by using one of the subcommands below."
}
func (f *FSCommand) Synopsis() string {
return "Inspect the contents of an allocation directory"
}
func (f *FSCommand) Run(args []string) int {
return cli.RunResultHelp
}

2
command/fs_cat.go
View File

@@ -29,7 +29,7 @@ func (f *FSCatCommand) Help() string {
}
func (f *FSCatCommand) Synopsis() string {
return "displays a file at a given location"
return "Cat a file in an allocation directory"
}
func (f *FSCatCommand) Run(args []string) int {

2
command/fs_ls.go
View File

@@ -28,7 +28,7 @@ Usage: nomad fs ls <alloc-id> <path>
}
func (f *FSListCommand) Synopsis() string {
return "Lists list of files of an allocation directory"
return "List files in an allocation directory"
}
func (f *FSListCommand) Run(args []string) int {

2
command/fs_stat.go
View File

@@ -27,7 +27,7 @@ Usage: nomad fs stat <alloc-id> <path>
}
func (f *FSStatCommand) Synopsis() string {
return "Stats an entry in an allocation directory"
return "Stat an entry in an allocation directory"
}
func (f *FSStatCommand) Run(args []string) int {

12
command/monitor.go
View File

@@ -299,12 +299,14 @@ func (m *monitor) monitor(evalID string, allowPrefix bool) int {
// Monitor the next eval in the chain, if present
if eval.NextEval != "" {
m.ui.Info(fmt.Sprintf(
"Monitoring next evaluation %q in %s",
eval.NextEval, eval.Wait))
if eval.Wait.Nanoseconds() != 0 {
m.ui.Info(fmt.Sprintf(
"Monitoring next evaluation %q in %s",
limit(eval.NextEval, m.length), eval.Wait))
// Skip some unnecessary polling
time.Sleep(eval.Wait)
// Skip some unnecessary polling
time.Sleep(eval.Wait)
}
// Reset the state and monitor the new eval
m.state = newEvalState()

5
commands.go
View File

@@ -62,6 +62,11 @@ func Commands(metaPtr *command.Meta) map[string]cli.CommandFactory {
Meta: meta,
}, nil
},
"fs": func() (cli.Command, error) {
return &command.FSCommand{
Meta: meta,
}, nil
},
"fs ls": func() (cli.Command, error) {
return &command.FSListCommand{
Meta: meta,

2
main.go Executable file → Normal file
View File

@@ -32,7 +32,7 @@ func RunCustom(args []string, commands map[string]cli.CommandFactory) int {
commandsInclude := make([]string, 0, len(commands))
for k, _ := range commands {
switch k {
case "spawn-daemon":
case "executor":
default:
commandsInclude = append(commandsInclude, k)
}

7
nomad/structs/structs.go
View File

@@ -1414,6 +1414,13 @@ func (s *Service) InitFields(job string, taskGroup string, task string) {
// Validate checks if the Check definition is valid
func (s *Service) Validate() error {
var mErr multierror.Error
// Ensure the name does not have a period in it.
// RFC-2782: https://tools.ietf.org/html/rfc2782
if strings.Contains(s.Name, ".") {
mErr.Errors = append(mErr.Errors, fmt.Errorf("service name can't contain periods: %q", s.Name))
}
for _, c := range s.Checks {
if err := c.Validate(); err != nil {
mErr.Errors = append(mErr.Errors, err)

8
nomad/structs/structs_test.go
View File

@@ -485,6 +485,14 @@ func TestInvalidServiceCheck(t *testing.T) {
if err := s.Validate(); err == nil {
t.Fatalf("Service should be invalid")
}
s = Service{
Name: "service.name",
PortLabel: "bar",
}
if err := s.Validate(); err == nil {
t.Fatalf("Service should be invalid: %v", err)
}
}
func TestDistinctCheckID(t *testing.T) {

2
testutil/wait.go
View File

@@ -11,7 +11,7 @@ import (
const (
// TravisRunEnv is an environment variable that is set if being run by
// Travis.
TravisRunEnv = "TRAVIS_RUN"
TravisRunEnv = "CI"
)
type testFn func() (bool, error)

3
website/source/docs/drivers/docker.html.md
View File

@@ -247,6 +247,9 @@ The `docker` driver has the following host-level configuration options:
to customize this if you use a non-standard socket (http or another
location).
* `docker.auth.config` - Allows an operator to specify a json file which is in
the dockercfg format containing authentication information for private registry.
* `docker.tls.cert` - Path to the server's certificate file (`.pem`). Specify
this along with `docker.tls.key` and `docker.tls.ca` to use a TLS client to
connect to the docker daemon. `docker.endpoint` must also be specified or

4
website/source/docs/http/node.html.md
View File

@@ -338,7 +338,7 @@ be specified using the `?region=` query parameter.
{
"EvalIDs": ["d092fdc0-e1fd-2536-67d8-43af8ca798ac"],
"EvalCreateIndex": 35,
"NodeModifyIndex": 34,
"NodeModifyIndex": 34
}
```
@@ -378,7 +378,7 @@ be specified using the `?region=` query parameter.
{
"EvalID": "d092fdc0-e1fd-2536-67d8-43af8ca798ac",
"EvalCreateIndex": 35,
"NodeModifyIndex": 34,
"NodeModifyIndex": 34
}
```

32
website/source/intro/getting-started/cluster.html.md
View File

@@ -140,9 +140,9 @@ we should see both nodes in the `ready` state:
```
$ nomad node-status
ID DC Name Class Drain Status
f7780117-2cae-8ee9-4b36-f34dd796ab02 dc1 nomad <none> false ready
ffb5b55a-6059-9ec7-6108-23a2bbba95da dc1 nomad <none> false ready
ID Datacenter Name Class Drain Status
fca62612 dc1 nomad <none> false ready
c887deef dc1 nomad <none> false ready
```
We now have a simple three node cluster running. The only difference
@@ -159,13 +159,13 @@ Then, use the [`run` command](/docs/commands/run.html) to submit the job:
```
$ nomad run example.nomad
==> Monitoring evaluation "77e5075f-2a1b-9cce-d14e-fe98cca9e17f"
==> Monitoring evaluation "8e0a7cf9"
Evaluation triggered by job "example"
Allocation "711edd85-f183-99ea-910a-6445b23d79e4" created: node "ffb5b55a-6059-9ec7-6108-23a2bbba95da", group "cache"
Allocation "98218a8a-627c-308f-8941-acdbffe1940c" created: node "f7780117-2cae-8ee9-4b36-f34dd796ab02", group "cache"
Allocation "e8957a7f-6fff-f61f-2878-57715c26725d" created: node "f7780117-2cae-8ee9-4b36-f34dd796ab02", group "cache"
Allocation "501154ac" created: node "c887deef", group "cache"
Allocation "7e2b3900" created: node "fca62612", group "cache"
Allocation "9c66fcaf" created: node "c887deef", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "77e5075f-2a1b-9cce-d14e-fe98cca9e17f" finished with status "complete"
==> Evaluation "8e0a7cf9" finished with status "complete"
```
We can see in the output that the scheduler assigned two of the
@@ -181,17 +181,19 @@ Name = example
Type = service
Priority = 50
Datacenters = dc1
Status = <none>
Status = running
Periodic = false
==> Evaluations
ID Priority TriggeredBy Status
77e5075f-2a1b-9cce-d14e-fe98cca9e17f 50 job-register complete
ID Priority Triggered By Status
54dd2ae3 50 job-register complete
==> Allocations
ID EvalID NodeID TaskGroup Desired Status
711edd85-f183-99ea-910a-6445b23d79e4 77e5075f-2a1b-9cce-d14e-fe98cca9e17f ffb5b55a-6059-9ec7-6108-23a2bbba95da cache run running
98218a8a-627c-308f-8941-acdbffe1940c 77e5075f-2a1b-9cce-d14e-fe98cca9e17f f7780117-2cae-8ee9-4b36-f34dd796ab02 cache run running
e8957a7f-6fff-f61f-2878-57715c26725d 77e5075f-2a1b-9cce-d14e-fe98cca9e17f f7780117-2cae-8ee9-4b36-f34dd796ab02 cache run running
ID Eval ID Node ID Task Group Desired Status
102225ab 54dd2ae3 56b590e6 cache run running
f327d2b1 54dd2ae3 e4235508 cache run running
f91137f8 54dd2ae3 56b590e6 cache run running
```
We can see that all our tasks have been allocated and are running.

1
website/source/intro/getting-started/install.html.md
View File

@@ -50,6 +50,7 @@ Available commands are:
alloc-status Display allocation status information and metadata
client-config View or modify client configuration details
eval-monitor Monitor an evaluation interactively
fs Inspect the contents of an allocation directory
init Create an example job file
node-drain Toggle drain mode on a given node
node-status Display status information about nodes

53
website/source/intro/getting-started/jobs.html.md
View File

@@ -46,11 +46,11 @@ We can register our example job now:
```
$ nomad run example.nomad
==> Monitoring evaluation "3d823c52-929a-fa8b-c50d-1ac4d00cf6b7"
==> Monitoring evaluation "26cfc69e"
Evaluation triggered by job "example"
Allocation "85b839d7-f67a-72a4-5a13-104020ae4807" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "8ba85cef" created: node "171a583b", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "3d823c52-929a-fa8b-c50d-1ac4d00cf6b7" finished with status "complete"
==> Evaluation "26cfc69e" finished with status "complete"
```
Anytime a job is updated, Nomad creates an evaluation to determine what
@@ -67,15 +67,16 @@ Name = example
Type = service
Priority = 50
Datacenters = dc1
Status = <none>
Status = running
Periodic = false
==> Evaluations
ID Priority TriggeredBy Status
3d823c52-929a-fa8b-c50d-1ac4d00cf6b7 50 job-register complete
ID Priority Triggered By Status
26cfc69e 50 job-register complete
==> Allocations
ID EvalID NodeID TaskGroup Desired Status
85b839d7-f67a-72a4-5a13-104020ae4807 3d823c52-929a-fa8b-c50d-1ac4d00cf6b7 2512929f-5b7c-a959-dfd9-bf8a8eb022a6 cache run running
ID Eval ID Node ID Task Group Desired Status
8ba85cef 26cfc69e 171a583b cache run running
```
Here we can see that our evaluation that was created has completed, and that
@@ -100,13 +101,13 @@ push the updated version of the job:
```
$ nomad run example.nomad
==> Monitoring evaluation "ec199c63-2022-f5c7-328d-1cf85e61bf66"
==> Monitoring evaluation "127a49d0"
Evaluation triggered by job "example"
Allocation "21551679-5224-cb6b-80a2-d0b091612d2e" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "b1be1410-a01c-20ad-80ff-96750ec0f1da" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "ed32a35d-8086-3f04-e299-4432e562cbf2" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "8ab24eef" created: node "171a583b", group "cache"
Allocation "f6c29874" created: node "171a583b", group "cache"
Allocation "8ba85cef" modified: node "171a583b", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "ec199c63-2022-f5c7-328d-1cf85e61bf66" finished with status "complete"
==> Evaluation "127a49d0" finished with status "complete"
```
Because we set the count of the task group to three, Nomad created two
@@ -132,13 +133,23 @@ specification now:
```
$ nomad run example.nomad
==> Monitoring evaluation "d34d37f4-19b1-f4c0-b2da-c949e6ade82d"
==> Monitoring evaluation "ebcc3e14"
Evaluation triggered by job "example"
Allocation "5614feb0-212d-21e5-ccfb-56a394fc41d5" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "bf7e3ad5-b217-14fe-f3f8-2b83af9dbb42" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "e3978af2-f61e-c601-7aa1-90aea9b23cf6" created: node "2512929f-5b7c-a959-dfd9-bf8a8eb022a6", group "cache"
Allocation "9a3743f4" created: node "171a583b", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "d34d37f4-19b1-f4c0-b2da-c949e6ade82d" finished with status "complete"
==> Evaluation "ebcc3e14" finished with status "complete"
==> Monitoring next evaluation "b508d8f0-7f21-8d66-ec59-7f5b2573435a" in 0
==> Monitoring evaluation "b508d8f0"
Evaluation triggered by job "example"
Allocation "926e5876" created: node "171a583b", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "b508d8f0" finished with status "complete"
==> Monitoring next evaluation "ea78c05a-a15f-92ae-8c3d-59f4a1edd091" in 10s
==> Monitoring evaluation "ea78c05a"
Evaluation triggered by job "example"
Allocation "3c8589d5" created: node "171a583b", group "cache"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "ea78c05a" finished with status "complete"
```
We can see that Nomad handled the update in three phases, only updating a single task
@@ -152,10 +163,10 @@ is stopping the job. This is done with the [`stop` command](/docs/commands/stop.
```
$ nomad stop example
==> Monitoring evaluation "bb407de4-02cb-f009-d986-646d6c11366d"
==> Monitoring evaluation "fd03c9f8"
Evaluation triggered by job "example"
Evaluation status changed: "pending" -> "complete"
==> Evaluation "bb407de4-02cb-f009-d986-646d6c11366d" finished with status "complete"
==> Evaluation "fd03c9f8" finished with status "complete"
```
When we stop a job, it creates an evaluation which is used to stop all
@@ -164,7 +175,7 @@ If we try to query the job status, we can see it is no longer registered:
```
$ nomad status example
Error querying job: Unexpected response code: 404 (job not found)
No job(s) with prefix or id "example" found
```
If we wanted to start the job again, we could simply `run` it again.

8
website/source/intro/getting-started/running.html.md
View File

@@ -83,8 +83,8 @@ $ vagrant ssh
...
$ nomad node-status
ID DC Name Class Drain Status
72d3af97-144f-1e5f-94e5-df1516fe4add dc1 nomad <none> false ready
ID Datacenter Name Class Drain Status
171a583b dc1 nomad <none> false ready
```
The output shows our Node ID, which is a randomly generated UUID,
@@ -99,8 +99,8 @@ ring using the [`server-members`](/docs/commands/server-members.html) command:
```text
$ nomad server-members
Name Addr Port Status Proto Build DC Region
nomad.global 127.0.0.1 4648 alive 2 0.1.0dev dc1 global
Name Address Port Status Protocol Build Datacenter Region
nomad.global 127.0.0.1 4648 alive 2 0.3.0dev dc1 global
```
The output shows our own agent, the address it is running on, its