Files
yms-daemon/internal/backendupdate/container.go
T

480 lines
21 KiB
Go
Raw Normal View History

2026-08-16 17:12:06 +08:00
package backendupdate
import (
"bytes"
"context"
"crypto/rand"
"encoding/json"
"errors"
"fmt"
"os"
"path/filepath"
"strings"
"github.com/distribution/reference"
"yms-daemon/internal/backendexecutor"
"yms-daemon/internal/containerengine"
"yms-daemon/internal/deploymentconfig"
"yms-daemon/internal/hostnginx"
"yms-daemon/internal/transaction"
)
const inputTypeContainerImage = "container-image"
type persistedContainerRequest struct {
InputType string `json:"inputType"`
ImageReference string `json:"imageReference"`
ImmutableReference string `json:"immutableReference"`
ImageDigest string `json:"imageDigest"`
Platform containerengine.Platform `json:"platform"`
TargetPort int `json:"targetPort"`
TargetContainer string `json:"targetContainer"`
PreviousPort int `json:"previousPort"`
PreviousContainer string `json:"previousContainer"`
TargetHealthEndpoint string `json:"targetHealthEndpoint"`
GatewayBeforePath string `json:"gatewayBeforePath"`
GatewayAfterPath string `json:"gatewayAfterPath"`
GatewayReceiptPath string `json:"gatewayReceiptPath"`
}
// UpdateContainerImage pulls one development image, freezes its repository
// digest, and updates the inactive Docker backend slot.
func (u *Updater) UpdateContainerImage(ctx context.Context, imageReference string, report ProgressReporter) (transaction.Transaction, error) {
if u.containerExecutor == nil || u.engine == nil {
return transaction.Transaction{}, errors.New("container backend updater is not configured")
}
if strings.TrimSpace(imageReference) != imageReference || imageReference == "" {
return transaction.Transaction{}, errors.New("exact container image reference is required")
}
configInfo, err := os.Lstat(u.containerConfigSource)
if err != nil {
return transaction.Transaction{}, fmt.Errorf("inspect backend configuration %s: %w", u.containerConfigSource, err)
}
if !configInfo.Mode().IsRegular() || configInfo.Mode()&os.ModeSymlink != 0 {
return transaction.Transaction{}, fmt.Errorf("backend configuration is not a direct regular file: %s", u.containerConfigSource)
}
tmpInfo, err := os.Lstat(u.containerTmpSource)
if err != nil {
return transaction.Transaction{}, fmt.Errorf("inspect backend temporary directory %s: %w", u.containerTmpSource, err)
}
if !tmpInfo.IsDir() || tmpInfo.Mode()&os.ModeSymlink != 0 {
return transaction.Transaction{}, fmt.Errorf("backend temporary path is not a direct directory: %s", u.containerTmpSource)
}
active, err := u.store.ActiveTransaction(ctx)
if err == nil {
var request persistedContainerRequest
if decodeErr := decodeContainerRequest(active.Request, &request); decodeErr != nil {
return active, decodeErr
}
if active.Service != serviceBackend || request.InputType != inputTypeContainerImage {
return active, &transaction.ActiveTransactionError{TransactionID: active.ID}
}
if request.ImageReference != imageReference {
return active, &transaction.ActiveTransactionError{TransactionID: active.ID}
}
return u.runContainerUpdate(ctx, active, request, false, report)
}
if !errors.Is(err, transaction.ErrNotFound) {
return transaction.Transaction{}, err
}
reportProgress(report, Progress{Message: "Pulling backend image " + imageReference})
resolved, err := u.pullAndResolveImage(ctx, imageReference)
if err != nil {
return transaction.Transaction{}, err
}
reportProgress(report, Progress{Message: "Backend image resolved: " + resolved.ImmutableReference})
before, err := u.gateway.Read()
if err != nil {
return transaction.Transaction{}, err
}
activeSlot, err := u.config.Backend.SlotForPort(before.ActivePort)
if err != nil {
return transaction.Transaction{}, err
}
deployment, deploymentErr := u.store.BackendContainerDeployment(ctx)
hasDeployment := deploymentErr == nil
if deploymentErr != nil && !errors.Is(deploymentErr, transaction.ErrNotFound) {
return transaction.Transaction{}, deploymentErr
}
hasHistory, err := u.store.HasCommittedBackendContainerTransactionHistory(ctx)
if err != nil {
return transaction.Transaction{}, err
}
targetPort, targetSlot, previousContainer, err := u.resolveContainerSlots(
ctx,
before.ActivePort,
activeSlot,
deployment,
hasDeployment,
hasHistory,
)
if err != nil {
return transaction.Transaction{}, err
}
afterContent, err := hostnginx.RenderBackendPort(before.Content, targetPort)
if err != nil {
return transaction.Transaction{}, err
}
transactionID := rand.Text()
transactionRoot := filepath.Join(u.workRoot, transactionID)
if err := os.MkdirAll(transactionRoot, 0o750); err != nil {
return transaction.Transaction{}, fmt.Errorf("create container backend transaction directory: %w", err)
}
beforePath := filepath.Join(transactionRoot, "gateway.before.conf")
afterPath := filepath.Join(transactionRoot, "gateway.after.conf")
if err := writeImmutableFile(beforePath, before.Content, 0o640); err != nil {
return transaction.Transaction{}, err
}
if err := writeImmutableFile(afterPath, afterContent, 0o640); err != nil {
return transaction.Transaction{}, err
}
request := persistedContainerRequest{
InputType: inputTypeContainerImage, ImageReference: imageReference,
ImmutableReference: resolved.ImmutableReference, ImageDigest: resolved.Digest, Platform: resolved.Platform,
TargetPort: targetPort, TargetContainer: targetSlot.ContainerName,
PreviousPort: before.ActivePort, PreviousContainer: previousContainer,
TargetHealthEndpoint: targetSlot.HealthEndpoint,
GatewayBeforePath: beforePath, GatewayAfterPath: afterPath,
GatewayReceiptPath: filepath.Join(transactionRoot, "gateway.applied"),
}
content, err := json.Marshal(request)
if err != nil {
return transaction.Transaction{}, fmt.Errorf("encode container backend update request: %w", err)
}
record, created, err := u.store.CreateTransaction(ctx, transaction.CreateRequest{
ID: transactionID, IdempotencyKey: serviceBackend + ":container:" + resolved.Digest,
Source: sourceLocalCLI, Service: serviceBackend, Request: content,
})
if err != nil {
return transaction.Transaction{}, err
}
if !created {
if err := decodeContainerRequest(record.Request, &request); err != nil {
return record, err
}
}
return u.runContainerUpdate(ctx, record, request, created, report)
}
// resolveContainerSlots reconciles the committed deployment, gateway and both
// exact container names before selecting a target. A fresh installation has no
// deployment row, no committed container transaction history and no slot
// containers. It starts on the non-routed slot so traffic is exposed only after
// health passes.
func (u *Updater) resolveContainerSlots(
ctx context.Context,
activePort int,
activeSlot deploymentconfig.BackendSlot,
deployment transaction.BackendContainerDeployment,
hasDeployment bool,
hasHistory bool,
) (int, deploymentconfig.BackendSlot, string, error) {
inactivePort := otherPort(activePort)
inactiveSlot, err := u.config.Backend.SlotForPort(inactivePort)
if err != nil {
return 0, deploymentconfig.BackendSlot{}, "", err
}
active, activeErr := u.engine.InspectContainer(ctx, activeSlot.ContainerName)
if activeErr != nil && !errors.Is(activeErr, containerengine.ErrNotFound) {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("inspect active backend container %s: %w", activeSlot.ContainerName, activeErr)
}
inactive, inactiveErr := u.engine.InspectContainer(ctx, inactiveSlot.ContainerName)
if inactiveErr != nil && !errors.Is(inactiveErr, containerengine.ErrNotFound) {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("inspect inactive backend container %s: %w", inactiveSlot.ContainerName, inactiveErr)
}
activeFound := activeErr == nil
inactiveFound := inactiveErr == nil
if hasDeployment {
if deployment.ActivePort != activePort {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("committed backend container port %d does not match gateway active port %d", deployment.ActivePort, activePort)
}
if deployment.ContainerName != activeSlot.ContainerName {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("committed backend container %s does not match gateway slot container %s", deployment.ContainerName, activeSlot.ContainerName)
}
if !activeFound {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("committed active backend container %s is missing", activeSlot.ContainerName)
}
if active.ID != deployment.ContainerID {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("active backend container %s identity does not match committed deployment", activeSlot.ContainerName)
}
}
if activeFound {
if !active.Running || active.Dead {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("active backend container %s is not running", activeSlot.ContainerName)
}
if inactiveFound && inactive.Running && !inactive.Dead {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("inactive backend container %s is unexpectedly running", inactiveSlot.ContainerName)
}
return inactivePort, inactiveSlot, activeSlot.ContainerName, nil
}
if hasHistory {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("active backend container %s is missing on a server with committed backend container transaction history", activeSlot.ContainerName)
}
if inactiveFound && inactive.Running && !inactive.Dead {
return 0, deploymentconfig.BackendSlot{}, "", fmt.Errorf("active backend container %s is missing but inactive container %s is running", activeSlot.ContainerName, inactiveSlot.ContainerName)
}
return inactivePort, inactiveSlot, "", nil
}
type resolvedImage struct {
ImmutableReference string
Digest string
Platform containerengine.Platform
}
func (u *Updater) pullAndResolveImage(ctx context.Context, imageReference string) (resolvedImage, error) {
named, err := reference.ParseNormalizedNamed(imageReference)
if err != nil {
return resolvedImage{}, fmt.Errorf("parse container image reference: %w", err)
}
if _, ok := named.(reference.Tagged); !ok {
return resolvedImage{}, errors.New("--container-image requires a tag-qualified image reference")
}
if err := u.engine.Ping(ctx); err != nil {
return resolvedImage{}, err
}
if err := u.engine.PullImage(ctx, imageReference); err != nil {
return resolvedImage{}, err
}
image, err := u.engine.InspectImage(ctx, imageReference)
if err != nil {
return resolvedImage{}, fmt.Errorf("inspect pulled backend image: %w", err)
}
if image.Platform.OS == "" || image.Platform.Architecture == "" {
return resolvedImage{}, errors.New("pulled backend image does not report an exact platform")
}
repository := reference.TrimNamed(named).Name()
matches := make(map[string]string)
for _, value := range image.RepoDigests {
digested, err := reference.ParseNormalizedNamed(value)
if err != nil {
return resolvedImage{}, fmt.Errorf("parse pulled repository digest %q: %w", value, err)
}
withDigest, ok := digested.(reference.Digested)
if !ok || reference.TrimNamed(digested).Name() != repository {
continue
}
matches[withDigest.Digest().String()] = value
}
if len(matches) == 0 {
return resolvedImage{}, fmt.Errorf("pulled backend image has no repository digest for %s", repository)
}
if len(matches) != 1 {
return resolvedImage{}, fmt.Errorf("pulled backend image has multiple repository digests for %s", repository)
}
for digest, immutableReference := range matches {
return resolvedImage{ImmutableReference: immutableReference, Digest: digest, Platform: image.Platform}, nil
}
return resolvedImage{}, errors.New("repository digest resolution produced no result")
}
func (u *Updater) runContainerUpdate(ctx context.Context, record transaction.Transaction, request persistedContainerRequest, created bool, report ProgressReporter) (transaction.Transaction, error) {
message := "Resuming container backend transaction " + record.ID
if created {
message = "Created container backend transaction " + record.ID
}
reportProgress(report, Progress{TransactionID: record.ID, State: record.State, Message: message})
if record.State.Terminal() {
return terminalResult(record)
}
executorRequest := backendexecutor.Request{
ImageAcquisition: backendexecutor.ImageAcquisitionPull,
ImageReference: request.ImmutableReference, ExpectedImageDigest: request.ImageDigest, Platform: request.Platform,
ContainerName: request.TargetContainer, Port: request.TargetPort,
PortEnvironmentKey: deploymentconfig.ContainerPortEnvironment,
ConfigSource: u.containerConfigSource, ConfigTarget: u.containerConfigTarget,
TmpSource: u.containerTmpSource, TmpTarget: u.containerTmpTarget,
ConfigEnvironmentKey: deploymentconfig.ContainerConfigEnvironment,
ConfigLocation: deploymentconfig.ContainerConfigLocation,
RestartPolicy: containerengine.RestartPolicy{Name: "no"},
HealthEndpoint: request.TargetHealthEndpoint,
}
switch record.State {
case transaction.StateCreated, transaction.StateValidating, transaction.StatePrepared, transaction.StateStarting:
if err := u.containerExecutor.Run(ctx, record.ID, executorRequest); err != nil {
current, readErr := u.store.Transaction(ctx, record.ID)
if readErr != nil || current.State == transaction.StateFailed {
return u.currentWithError(ctx, record.ID, errors.Join(err, readErr))
}
before, beforeErr := readGatewaySnapshot(request.GatewayBeforePath, request.PreviousPort)
after, afterErr := readGatewaySnapshot(request.GatewayAfterPath, request.TargetPort)
if snapshotErr := errors.Join(beforeErr, afterErr); snapshotErr != nil {
return u.currentWithError(ctx, record.ID, errors.Join(err, snapshotErr))
}
rollbackErr := u.rollbackContainer(ctx, record.ID, request, before, after, err)
return u.currentWithError(ctx, record.ID, rollbackErr)
}
case transaction.StateSwitching, transaction.StateVerifying, transaction.StateDraining, transaction.StateRollingBack:
default:
return u.currentWithError(ctx, record.ID, fmt.Errorf("container backend update cannot resume transaction %s in state %s", record.ID, record.State))
}
if err := u.switchAndCommitContainer(ctx, record.ID, request, report); err != nil {
return u.currentWithError(ctx, record.ID, err)
}
return u.store.Transaction(ctx, record.ID)
}
func (u *Updater) switchAndCommitContainer(ctx context.Context, transactionID string, request persistedContainerRequest, report ProgressReporter) error {
before, err := readGatewaySnapshot(request.GatewayBeforePath, request.PreviousPort)
if err != nil {
return err
}
after, err := readGatewaySnapshot(request.GatewayAfterPath, request.TargetPort)
if err != nil {
return err
}
for {
record, err := u.store.Transaction(ctx, transactionID)
if err != nil {
return err
}
switch record.State {
case transaction.StateSwitching:
reportProgress(report, Progress{TransactionID: transactionID, State: record.State, Message: fmt.Sprintf("Switching host Nginx backend traffic to port %d", request.TargetPort)})
operation := &gatewayOperation{controller: u.gateway, before: before, after: after, receiptPath: request.GatewayReceiptPath}
if _, err := u.coordinator.ExecuteStep(ctx, transactionID, containerGatewaySwitchIntent(request), operation); err != nil {
return u.rollbackContainer(ctx, transactionID, request, before, after, err)
}
if _, err := u.store.Transition(ctx, transactionID, transaction.StateVerifying, "host Nginx now routes backend traffic to the healthy container slot"); err != nil {
return err
}
case transaction.StateVerifying:
message := "container backend switch verified"
if request.PreviousContainer != "" {
message += "; previous container draining"
}
if _, err := u.store.Transition(ctx, transactionID, transaction.StateDraining, message); err != nil {
return err
}
case transaction.StateDraining:
if request.PreviousContainer != "" {
reportProgress(report, Progress{TransactionID: transactionID, State: record.State, Message: fmt.Sprintf("Draining previous backend container for %s", u.drain)})
if err := waitContext(ctx, u.drain); err != nil {
return err
}
operation := &containerStopOperation{engine: u.engine, name: request.PreviousContainer}
if _, err := u.coordinator.ExecuteStep(ctx, transactionID, stopPreviousContainerIntent(request), operation); err != nil {
return err
}
}
target, err := u.engine.InspectContainer(ctx, request.TargetContainer)
if err != nil {
return fmt.Errorf("inspect target backend container before commit: %w", err)
}
if !target.Running || target.Dead || target.ID == "" {
return fmt.Errorf("target backend container %s is not running with an exact identity", request.TargetContainer)
}
if target.ImageReference != request.ImmutableReference {
return fmt.Errorf("target backend container %s image reference does not match the transaction", request.TargetContainer)
}
_, err = u.store.CommitBackendContainerDeployment(ctx, transactionID, transaction.BackendContainerDeployment{
ActivePort: request.TargetPort,
ContainerName: request.TargetContainer,
ImageDigest: request.ImageDigest,
ContainerID: target.ID,
}, "container backend update committed")
if err == nil {
reportProgress(report, Progress{TransactionID: transactionID, State: transaction.StateCommitted, Message: "Container backend update committed"})
}
return err
case transaction.StateCommitted:
return nil
case transaction.StateRollingBack:
return u.rollbackContainer(ctx, transactionID, request, before, after, errors.New("resuming container backend rollback"))
default:
return fmt.Errorf("container backend commit cannot continue transaction %s in state %s", transactionID, record.State)
}
}
}
func (u *Updater) rollbackContainer(ctx context.Context, transactionID string, request persistedContainerRequest, before hostnginx.Snapshot, after hostnginx.Snapshot, cause error) error {
record, err := u.store.Transaction(ctx, transactionID)
if err != nil {
return errors.Join(cause, err)
}
if record.State != transaction.StateRollingBack {
if _, err := u.store.Transition(ctx, transactionID, transaction.StateRollingBack, "container backend compensation started"); err != nil {
return errors.Join(cause, err)
}
}
gateway := &gatewayOperation{controller: u.gateway, before: after, after: before, receiptPath: request.GatewayReceiptPath + ".restore"}
_, gatewayErr := u.coordinator.ExecuteStep(ctx, transactionID, containerGatewayRestoreIntent(request), gateway)
stop := &containerStopOperation{engine: u.engine, name: request.TargetContainer}
_, stopErr := u.coordinator.ExecuteStep(ctx, transactionID, stopTargetContainerIntent(request), stop)
if err := errors.Join(gatewayErr, stopErr); err != nil {
return errors.Join(cause, err)
}
_, transitionErr := u.store.Transition(ctx, transactionID, transaction.StateRolledBack, "container backend compensation completed")
return errors.Join(cause, transitionErr)
}
type containerStopOperation struct {
engine containerengine.Engine
name string
}
func (o *containerStopOperation) Apply(ctx context.Context) error {
err := o.engine.StopContainer(ctx, o.name)
if errors.Is(err, containerengine.ErrNotFound) {
return nil
}
return err
}
func (o *containerStopOperation) Inspect(ctx context.Context) (transaction.Inspection, error) {
record, err := o.engine.InspectContainer(ctx, o.name)
if errors.Is(err, containerengine.ErrNotFound) {
return transaction.Inspection{Status: transaction.InspectionApplied}, nil
}
if err != nil {
return transaction.Inspection{}, err
}
result, _ := json.Marshal(record)
if !record.Running || record.Dead {
return transaction.Inspection{Status: transaction.InspectionApplied, Result: result}, nil
}
return transaction.Inspection{Status: transaction.InspectionNotApplied, Result: result}, nil
}
func containerGatewaySwitchIntent(request persistedContainerRequest) transaction.StepIntent {
return stepIntent("backend.container.gateway.switch", "switch host Nginx to container backend", struct {
BeforePort int `json:"beforePort"`
AfterPort int `json:"afterPort"`
}{request.PreviousPort, request.TargetPort})
}
func containerGatewayRestoreIntent(request persistedContainerRequest) transaction.StepIntent {
return stepIntent("backend.container.gateway.restore", "restore host Nginx after container backend failure", struct {
Port int `json:"port"`
}{request.PreviousPort})
}
func stopPreviousContainerIntent(request persistedContainerRequest) transaction.StepIntent {
return stepIntent("backend.previous-container.stop", "stop previous backend container after drain", struct {
ContainerName string `json:"containerName"`
}{request.PreviousContainer})
}
func stopTargetContainerIntent(request persistedContainerRequest) transaction.StepIntent {
return stepIntent("backend.target-container.stop", "stop compensated backend container", struct {
ContainerName string `json:"containerName"`
}{request.TargetContainer})
}
func decodeContainerRequest(content json.RawMessage, request *persistedContainerRequest) error {
decoder := json.NewDecoder(bytes.NewReader(content))
decoder.DisallowUnknownFields()
if err := decoder.Decode(request); err != nil {
return fmt.Errorf("decode persisted container backend request: %w", err)
}
return nil
}
var _ transaction.Operation = (*containerStopOperation)(nil)