open-nomad/client/allocrunner/taskrunner/envoy_version_hook_test.go

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

477 lines
14 KiB
Go
Raw Normal View History

// Copyright (c) HashiCorp, Inc.
// SPDX-License-Identifier: MPL-2.0
package taskrunner
import (
"context"
2022-04-02 00:24:02 +00:00
"errors"
"testing"
"github.com/hashicorp/nomad/ci"
"github.com/hashicorp/nomad/client/allocdir"
ifs "github.com/hashicorp/nomad/client/allocrunner/interfaces"
"github.com/hashicorp/nomad/client/taskenv"
"github.com/hashicorp/nomad/command/agent/consul"
"github.com/hashicorp/nomad/helper/envoy"
"github.com/hashicorp/nomad/helper/testlog"
"github.com/hashicorp/nomad/nomad/mock"
"github.com/hashicorp/nomad/nomad/structs"
"github.com/shoenig/test/must"
)
var (
taskEnvDefault = taskenv.NewTaskEnv(nil, nil, nil, map[string]string{
"meta.connect.sidecar_image": envoy.ImageFormat,
"meta.connect.gateway_image": envoy.ImageFormat,
}, "", "")
)
func TestEnvoyVersionHook_semver(t *testing.T) {
ci.Parallel(t)
t.Run("with v", func(t *testing.T) {
result, err := semver("v1.2.3")
must.NoError(t, err)
must.Eq(t, "1.2.3", result)
})
t.Run("without v", func(t *testing.T) {
result, err := semver("1.2.3")
must.NoError(t, err)
must.Eq(t, "1.2.3", result)
})
t.Run("unexpected", func(t *testing.T) {
_, err := semver("foo")
must.ErrorContains(t, err, "unexpected envoy version format: Malformed version: foo")
})
}
func TestEnvoyVersionHook_taskImage(t *testing.T) {
ci.Parallel(t)
t.Run("absent", func(t *testing.T) {
result := (*envoyVersionHook)(nil).taskImage(map[string]interface{}{
// empty
})
must.Eq(t, envoy.ImageFormat, result)
})
t.Run("not a string", func(t *testing.T) {
result := (*envoyVersionHook)(nil).taskImage(map[string]interface{}{
"image": 7, // not a string
})
must.Eq(t, envoy.ImageFormat, result)
})
t.Run("normal", func(t *testing.T) {
result := (*envoyVersionHook)(nil).taskImage(map[string]interface{}{
"image": "custom/envoy:latest",
})
must.Eq(t, "custom/envoy:latest", result)
})
}
func TestEnvoyVersionHook_tweakImage(t *testing.T) {
ci.Parallel(t)
image := envoy.ImageFormat
t.Run("legacy", func(t *testing.T) {
result, err := (*envoyVersionHook)(nil).tweakImage(image, nil)
must.NoError(t, err)
must.Eq(t, envoy.FallbackImage, result)
})
t.Run("unexpected", func(t *testing.T) {
_, err := (*envoyVersionHook)(nil).tweakImage(image, map[string][]string{
"envoy": {"foo", "bar", "baz"},
})
must.ErrorContains(t, err, "unexpected envoy version format: Malformed version: foo")
})
t.Run("standard envoy", func(t *testing.T) {
result, err := (*envoyVersionHook)(nil).tweakImage(image, map[string][]string{
"envoy": {"1.15.0", "1.14.4", "1.13.4", "1.12.6"},
})
must.NoError(t, err)
must.Eq(t, "docker.io/envoyproxy/envoy:v1.15.0", result)
})
t.Run("custom image", func(t *testing.T) {
custom := "custom-${NOMAD_envoy_version}/envoy:${NOMAD_envoy_version}"
result, err := (*envoyVersionHook)(nil).tweakImage(custom, map[string][]string{
"envoy": {"1.15.0", "1.14.4", "1.13.4", "1.12.6"},
})
must.NoError(t, err)
must.Eq(t, "custom-1.15.0/envoy:1.15.0", result)
})
}
func TestEnvoyVersionHook_interpolateImage(t *testing.T) {
ci.Parallel(t)
hook := (*envoyVersionHook)(nil)
t.Run("default sidecar", func(t *testing.T) {
task := &structs.Task{
Config: map[string]interface{}{"image": envoy.SidecarConfigVar},
}
hook.interpolateImage(task, taskEnvDefault)
must.Eq(t, envoy.ImageFormat, task.Config["image"])
})
t.Run("default gateway", func(t *testing.T) {
task := &structs.Task{
Config: map[string]interface{}{"image": envoy.GatewayConfigVar},
}
hook.interpolateImage(task, taskEnvDefault)
must.Eq(t, envoy.ImageFormat, task.Config["image"])
})
t.Run("custom static", func(t *testing.T) {
task := &structs.Task{
Config: map[string]interface{}{"image": "custom/envoy"},
}
hook.interpolateImage(task, taskEnvDefault)
must.Eq(t, "custom/envoy", task.Config["image"])
})
t.Run("custom interpolated", func(t *testing.T) {
task := &structs.Task{
Config: map[string]interface{}{"image": "${MY_ENVOY}"},
}
hook.interpolateImage(task, taskenv.NewTaskEnv(map[string]string{
"MY_ENVOY": "my/envoy",
}, map[string]string{
"MY_ENVOY": "my/envoy",
}, nil, nil, "", ""))
must.Eq(t, "my/envoy", task.Config["image"])
})
t.Run("no image", func(t *testing.T) {
task := &structs.Task{
Config: map[string]interface{}{},
}
hook.interpolateImage(task, taskEnvDefault)
must.MapEmpty(t, task.Config)
})
}
func TestEnvoyVersionHook_skip(t *testing.T) {
ci.Parallel(t)
h := new(envoyVersionHook)
t.Run("not docker", func(t *testing.T) {
skip := h.skip(&ifs.TaskPrestartRequest{
Task: &structs.Task{
Driver: "exec",
Config: nil,
},
})
must.True(t, skip)
})
t.Run("not connect", func(t *testing.T) {
skip := h.skip(&ifs.TaskPrestartRequest{
Task: &structs.Task{
Driver: "docker",
Kind: "",
},
})
must.True(t, skip)
})
t.Run("version not needed", func(t *testing.T) {
skip := h.skip(&ifs.TaskPrestartRequest{
Task: &structs.Task{
Driver: "docker",
Kind: structs.NewTaskKind(structs.ConnectProxyPrefix, "task"),
Config: map[string]interface{}{
"image": "custom/envoy:latest",
},
},
})
must.True(t, skip)
})
t.Run("version needed custom", func(t *testing.T) {
skip := h.skip(&ifs.TaskPrestartRequest{
Task: &structs.Task{
Driver: "docker",
Kind: structs.NewTaskKind(structs.ConnectProxyPrefix, "task"),
Config: map[string]interface{}{
"image": "custom/envoy:v${NOMAD_envoy_version}",
},
},
})
must.False(t, skip)
})
t.Run("version needed standard", func(t *testing.T) {
skip := h.skip(&ifs.TaskPrestartRequest{
Task: &structs.Task{
Driver: "docker",
Kind: structs.NewTaskKind(structs.ConnectProxyPrefix, "task"),
Config: map[string]interface{}{
"image": envoy.ImageFormat,
},
},
})
must.False(t, skip)
})
}
func TestTaskRunner_EnvoyVersionHook_Prestart_standard(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Setup a mock for Consul API
spAPI := consul.MockSupportedProxiesAPI{
Value: map[string][]string{
"envoy": {"1.15.0", "1.14.4"},
},
Error: nil,
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, spAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook
must.NoError(t, h.Prestart(context.Background(), request, &response))
// Assert the Task.Config[image] is concrete
must.Eq(t, "docker.io/envoyproxy/envoy:v1.15.0", request.Task.Config["image"])
}
func TestTaskRunner_EnvoyVersionHook_Prestart_custom(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
alloc.Job.TaskGroups[0].Tasks[0].Driver = "podman"
alloc.Job.TaskGroups[0].Tasks[0].Config["image"] = "custom-${NOMAD_envoy_version}:latest"
client: never embed alloc_dir in chroot Fixes #2522 Skip embedding client.alloc_dir when building chroot. If a user configures a Nomad client agent so that the chroot_env will embed the client.alloc_dir, Nomad will happily infinitely recurse while building the chroot until something horrible happens. The best case scenario is the filesystem's path length limit is hit. The worst case scenario is disk space is exhausted. A bad agent configuration will look something like this: ```hcl data_dir = "/tmp/nomad-badagent" client { enabled = true chroot_env { # Note that the source matches the data_dir "/tmp/nomad-badagent" = "/ohno" # ... } } ``` Note that `/ohno/client` (the state_dir) will still be created but not `/ohno/alloc` (the alloc_dir). While I cannot think of a good reason why someone would want to embed Nomad's client (and possibly server) directories in chroots, there should be no cause for harm. chroots are only built when Nomad runs as root, and Nomad disables running exec jobs as root by default. Therefore even if client state is copied into chroots, it will be inaccessible to tasks. Skipping the `data_dir` and `{client,server}.state_dir` is possible, but this PR attempts to implement the minimum viable solution to reduce risk of unintended side effects or bugs. When running tests as root in a vm without the fix, the following error occurs: ``` === RUN TestAllocDir_SkipAllocDir alloc_dir_test.go:520: Error Trace: alloc_dir_test.go:520 Error: Received unexpected error: Couldn't create destination file /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/testtask/nomad/test/testtask/.../nomad/test/testtask/secrets/.nomad-mount: open /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/.../testtask/secrets/.nomad-mount: file name too long Test: TestAllocDir_SkipAllocDir --- FAIL: TestAllocDir_SkipAllocDir (22.76s) ``` Also removed unused Copy methods on AllocDir and TaskDir structs. Thanks to @eveld for not letting me forget about this!
2021-10-15 23:56:14 +00:00
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Setup a mock for Consul API
spAPI := consul.MockSupportedProxiesAPI{
Value: map[string][]string{
"envoy": {"1.14.1", "1.13.3"},
},
Error: nil,
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, spAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook
must.NoError(t, h.Prestart(context.Background(), request, &response))
// Assert the Task.Config[image] is concrete
must.Eq(t, "custom-1.14.1:latest", request.Task.Config["image"])
}
func TestTaskRunner_EnvoyVersionHook_Prestart_skip(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
alloc.Job.TaskGroups[0].Tasks[0].Driver = "exec"
alloc.Job.TaskGroups[0].Tasks[0].Config = map[string]interface{}{
"command": "/sidecar",
}
client: never embed alloc_dir in chroot Fixes #2522 Skip embedding client.alloc_dir when building chroot. If a user configures a Nomad client agent so that the chroot_env will embed the client.alloc_dir, Nomad will happily infinitely recurse while building the chroot until something horrible happens. The best case scenario is the filesystem's path length limit is hit. The worst case scenario is disk space is exhausted. A bad agent configuration will look something like this: ```hcl data_dir = "/tmp/nomad-badagent" client { enabled = true chroot_env { # Note that the source matches the data_dir "/tmp/nomad-badagent" = "/ohno" # ... } } ``` Note that `/ohno/client` (the state_dir) will still be created but not `/ohno/alloc` (the alloc_dir). While I cannot think of a good reason why someone would want to embed Nomad's client (and possibly server) directories in chroots, there should be no cause for harm. chroots are only built when Nomad runs as root, and Nomad disables running exec jobs as root by default. Therefore even if client state is copied into chroots, it will be inaccessible to tasks. Skipping the `data_dir` and `{client,server}.state_dir` is possible, but this PR attempts to implement the minimum viable solution to reduce risk of unintended side effects or bugs. When running tests as root in a vm without the fix, the following error occurs: ``` === RUN TestAllocDir_SkipAllocDir alloc_dir_test.go:520: Error Trace: alloc_dir_test.go:520 Error: Received unexpected error: Couldn't create destination file /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/testtask/nomad/test/testtask/.../nomad/test/testtask/secrets/.nomad-mount: open /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/.../testtask/secrets/.nomad-mount: file name too long Test: TestAllocDir_SkipAllocDir --- FAIL: TestAllocDir_SkipAllocDir (22.76s) ``` Also removed unused Copy methods on AllocDir and TaskDir structs. Thanks to @eveld for not letting me forget about this!
2021-10-15 23:56:14 +00:00
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Setup a mock for Consul API
spAPI := consul.MockSupportedProxiesAPI{
Value: map[string][]string{
"envoy": {"1.14.1", "1.13.3"},
},
Error: nil,
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, spAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook
must.NoError(t, h.Prestart(context.Background(), request, &response))
// Assert the Task.Config[image] does not get set
must.MapNotContainsKey(t, request.Task.Config, "image")
}
func TestTaskRunner_EnvoyVersionHook_Prestart_fallback(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
client: never embed alloc_dir in chroot Fixes #2522 Skip embedding client.alloc_dir when building chroot. If a user configures a Nomad client agent so that the chroot_env will embed the client.alloc_dir, Nomad will happily infinitely recurse while building the chroot until something horrible happens. The best case scenario is the filesystem's path length limit is hit. The worst case scenario is disk space is exhausted. A bad agent configuration will look something like this: ```hcl data_dir = "/tmp/nomad-badagent" client { enabled = true chroot_env { # Note that the source matches the data_dir "/tmp/nomad-badagent" = "/ohno" # ... } } ``` Note that `/ohno/client` (the state_dir) will still be created but not `/ohno/alloc` (the alloc_dir). While I cannot think of a good reason why someone would want to embed Nomad's client (and possibly server) directories in chroots, there should be no cause for harm. chroots are only built when Nomad runs as root, and Nomad disables running exec jobs as root by default. Therefore even if client state is copied into chroots, it will be inaccessible to tasks. Skipping the `data_dir` and `{client,server}.state_dir` is possible, but this PR attempts to implement the minimum viable solution to reduce risk of unintended side effects or bugs. When running tests as root in a vm without the fix, the following error occurs: ``` === RUN TestAllocDir_SkipAllocDir alloc_dir_test.go:520: Error Trace: alloc_dir_test.go:520 Error: Received unexpected error: Couldn't create destination file /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/testtask/nomad/test/testtask/.../nomad/test/testtask/secrets/.nomad-mount: open /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/.../testtask/secrets/.nomad-mount: file name too long Test: TestAllocDir_SkipAllocDir --- FAIL: TestAllocDir_SkipAllocDir (22.76s) ``` Also removed unused Copy methods on AllocDir and TaskDir structs. Thanks to @eveld for not letting me forget about this!
2021-10-15 23:56:14 +00:00
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Setup a mock for Consul API
spAPI := consul.MockSupportedProxiesAPI{
Value: nil, // old consul, no .xDS.SupportedProxies
Error: nil,
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, spAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook
must.NoError(t, h.Prestart(context.Background(), request, &response))
// Assert the Task.Config[image] is the fallback image
must.Eq(t, "envoyproxy/envoy:v1.11.2@sha256:a7769160c9c1a55bb8d07a3b71ce5d64f72b1f665f10d81aa1581bc3cf850d09", request.Task.Config["image"])
}
func TestTaskRunner_EnvoyVersionHook_Prestart_error(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
client: never embed alloc_dir in chroot Fixes #2522 Skip embedding client.alloc_dir when building chroot. If a user configures a Nomad client agent so that the chroot_env will embed the client.alloc_dir, Nomad will happily infinitely recurse while building the chroot until something horrible happens. The best case scenario is the filesystem's path length limit is hit. The worst case scenario is disk space is exhausted. A bad agent configuration will look something like this: ```hcl data_dir = "/tmp/nomad-badagent" client { enabled = true chroot_env { # Note that the source matches the data_dir "/tmp/nomad-badagent" = "/ohno" # ... } } ``` Note that `/ohno/client` (the state_dir) will still be created but not `/ohno/alloc` (the alloc_dir). While I cannot think of a good reason why someone would want to embed Nomad's client (and possibly server) directories in chroots, there should be no cause for harm. chroots are only built when Nomad runs as root, and Nomad disables running exec jobs as root by default. Therefore even if client state is copied into chroots, it will be inaccessible to tasks. Skipping the `data_dir` and `{client,server}.state_dir` is possible, but this PR attempts to implement the minimum viable solution to reduce risk of unintended side effects or bugs. When running tests as root in a vm without the fix, the following error occurs: ``` === RUN TestAllocDir_SkipAllocDir alloc_dir_test.go:520: Error Trace: alloc_dir_test.go:520 Error: Received unexpected error: Couldn't create destination file /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/testtask/nomad/test/testtask/.../nomad/test/testtask/secrets/.nomad-mount: open /tmp/TestAllocDir_SkipAllocDir1457747331/001/nomad/test/.../testtask/secrets/.nomad-mount: file name too long Test: TestAllocDir_SkipAllocDir --- FAIL: TestAllocDir_SkipAllocDir (22.76s) ``` Also removed unused Copy methods on AllocDir and TaskDir structs. Thanks to @eveld for not letting me forget about this!
2021-10-15 23:56:14 +00:00
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Setup a mock for Consul API
spAPI := consul.MockSupportedProxiesAPI{
Value: nil,
Error: errors.New("some consul error"),
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, spAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook, error should be recoverable
err := h.Prestart(context.Background(), request, &response)
must.ErrorContains(t, err, "error retrieving supported Envoy versions from Consul: some consul error")
}
func TestTaskRunner_EnvoyVersionHook_Prestart_restart(t *testing.T) {
ci.Parallel(t)
logger := testlog.HCLogger(t)
// Setup an Allocation
alloc := mock.ConnectAlloc()
alloc.Job.TaskGroups[0].Tasks[0] = mock.ConnectSidecarTask()
allocDir, cleanupDir := allocdir.TestAllocDir(t, logger, "EnvoyVersionHook", alloc.ID)
defer cleanupDir()
// Set up a mock for Consul API.
mockProxiesAPI := consul.MockSupportedProxiesAPI{
Value: map[string][]string{
"envoy": {"1.15.0", "1.14.4"},
},
Error: nil,
}
// Run envoy_version hook
h := newEnvoyVersionHook(newEnvoyVersionHookConfig(alloc, mockProxiesAPI, logger))
// Create a prestart request
request := &ifs.TaskPrestartRequest{
Task: alloc.Job.TaskGroups[0].Tasks[0],
TaskDir: allocDir.NewTaskDir(alloc.Job.TaskGroups[0].Tasks[0].Name),
TaskEnv: taskEnvDefault,
}
must.NoError(t, request.TaskDir.Build(false, nil))
// Prepare a response
var response ifs.TaskPrestartResponse
// Run the hook and ensure the tasks image has been modified.
must.NoError(t, h.Prestart(context.Background(), request, &response))
must.Eq(t, "docker.io/envoyproxy/envoy:v1.15.0", request.Task.Config["image"])
// Overwrite the previously modified image. This is the same behaviour that
// occurs when the server sends a non-destructive allocation update.
request.Task.Config["image"] = "${meta.connect.sidecar_image}"
// Run the Prestart hook function again, and ensure the image is updated.
must.NoError(t, h.Prestart(context.Background(), request, &response))
must.Eq(t, "docker.io/envoyproxy/envoy:v1.15.0", request.Task.Config["image"])
// Run the hook again, and ensure the config is still the same mimicking
// a non-user initiated restart.
must.NoError(t, h.Prestart(context.Background(), request, &response))
must.Eq(t, "docker.io/envoyproxy/envoy:v1.15.0", request.Task.Config["image"])
}