mirror of https://github.com/docker/compose.git
385 lines
10 KiB
Go
385 lines
10 KiB
Go
/*
|
|
Copyright 2024 Docker Compose CLI authors
|
|
|
|
Licensed under the Apache License, Version 2.0 (the "License");
|
|
you may not use this file except in compliance with the License.
|
|
You may obtain a copy of the License at
|
|
|
|
http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
Unless required by applicable law or agreed to in writing, software
|
|
distributed under the License is distributed on an "AS IS" BASIS,
|
|
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
See the License for the specific language governing permissions and
|
|
limitations under the License.
|
|
*/
|
|
|
|
package desktop
|
|
|
|
import (
|
|
"context"
|
|
"errors"
|
|
"fmt"
|
|
"strings"
|
|
"sync"
|
|
|
|
"github.com/docker/compose/v2/internal/paths"
|
|
"github.com/docker/compose/v2/pkg/api"
|
|
"github.com/docker/compose/v2/pkg/progress"
|
|
"github.com/docker/go-units"
|
|
"github.com/sirupsen/logrus"
|
|
)
|
|
|
|
// fileShareProgressID is the identifier used for the root grouping of file
|
|
// share events in the progress writer.
|
|
const fileShareProgressID = "Synchronized File Shares"
|
|
|
|
// RemoveFileSharesForProject removes any Synchronized File Shares that were
|
|
// created by Compose for this project in the past if possible.
|
|
//
|
|
// Errors are not propagated; they are only sent to the progress writer.
|
|
func RemoveFileSharesForProject(ctx context.Context, c *Client, projectName string) {
|
|
w := progress.ContextWriter(ctx)
|
|
|
|
existing, err := c.ListFileShares(ctx)
|
|
if err != nil {
|
|
w.TailMsgf("Synchronized File Shares not removed due to error: %v", err)
|
|
return
|
|
}
|
|
// filter the list first, so we can early return and not show the event if
|
|
// there's no sessions to clean up
|
|
var toRemove []FileShareSession
|
|
for _, share := range existing {
|
|
if share.Labels["com.docker.compose.project"] == projectName {
|
|
toRemove = append(toRemove, share)
|
|
}
|
|
}
|
|
if len(toRemove) == 0 {
|
|
return
|
|
}
|
|
|
|
w.Event(progress.NewEvent(fileShareProgressID, progress.Working, "Removing"))
|
|
rootResult := progress.Done
|
|
defer func() {
|
|
w.Event(progress.NewEvent(fileShareProgressID, rootResult, ""))
|
|
}()
|
|
for _, share := range toRemove {
|
|
shareID := share.Labels["com.docker.desktop.mutagen.file-share.id"]
|
|
if shareID == "" {
|
|
w.Event(progress.Event{
|
|
ID: share.Alpha.Path,
|
|
ParentID: fileShareProgressID,
|
|
Status: progress.Warning,
|
|
StatusText: "Invalid",
|
|
})
|
|
continue
|
|
}
|
|
|
|
w.Event(progress.Event{
|
|
ID: share.Alpha.Path,
|
|
ParentID: fileShareProgressID,
|
|
Status: progress.Working,
|
|
})
|
|
|
|
var status progress.EventStatus
|
|
var statusText string
|
|
if err := c.DeleteFileShare(ctx, shareID); err != nil {
|
|
// TODO(milas): Docker Desktop is doing weird things with error responses,
|
|
// once fixed, we can return proper error types from the client
|
|
if strings.Contains(err.Error(), "file share in use") {
|
|
status = progress.Warning
|
|
statusText = "Resource is still in use"
|
|
if rootResult != progress.Error {
|
|
// error takes precedence over warning
|
|
rootResult = progress.Warning
|
|
}
|
|
} else {
|
|
logrus.Debugf("Error deleting file share %q: %v", shareID, err)
|
|
status = progress.Error
|
|
rootResult = progress.Error
|
|
}
|
|
} else {
|
|
logrus.Debugf("Deleted file share: %s", shareID)
|
|
status = progress.Done
|
|
}
|
|
|
|
w.Event(progress.Event{
|
|
ID: share.Alpha.Path,
|
|
ParentID: fileShareProgressID,
|
|
Status: status,
|
|
StatusText: statusText,
|
|
})
|
|
}
|
|
}
|
|
|
|
// FileShareManager maps between Compose bind mounts and Desktop File Shares
|
|
// state.
|
|
type FileShareManager struct {
|
|
mu sync.Mutex
|
|
cli *Client
|
|
projectName string
|
|
hostPaths []string
|
|
// state holds session status keyed by file share ID.
|
|
state map[string]*FileShareSession
|
|
}
|
|
|
|
func NewFileShareManager(cli *Client, projectName string, hostPaths []string) *FileShareManager {
|
|
return &FileShareManager{
|
|
cli: cli,
|
|
projectName: projectName,
|
|
hostPaths: hostPaths,
|
|
state: make(map[string]*FileShareSession),
|
|
}
|
|
}
|
|
|
|
// EnsureExists looks for existing File Shares or creates new ones for the
|
|
// host paths.
|
|
//
|
|
// This function blocks until each share reaches steady state, at which point
|
|
// flow can continue.
|
|
func (m *FileShareManager) EnsureExists(ctx context.Context) (err error) {
|
|
w := progress.ContextWriter(ctx)
|
|
w.Event(progress.NewEvent(fileShareProgressID, progress.Working, ""))
|
|
defer func() {
|
|
if err != nil {
|
|
w.Event(progress.NewEvent(fileShareProgressID, progress.Error, ""))
|
|
} else {
|
|
w.Event(progress.NewEvent(fileShareProgressID, progress.Done, ""))
|
|
}
|
|
}()
|
|
|
|
wait := &waiter{
|
|
shareIDs: make(map[string]struct{}),
|
|
done: make(chan struct{}),
|
|
}
|
|
|
|
handler := m.eventHandler(w, wait)
|
|
|
|
ctx, cancel := context.WithCancel(ctx)
|
|
defer cancel()
|
|
|
|
// stream session events to update internal state for project
|
|
monitorErr := make(chan error, 1)
|
|
go func() {
|
|
defer close(monitorErr)
|
|
if err := m.watch(ctx, handler); err != nil && ctx.Err() == nil {
|
|
monitorErr <- err
|
|
}
|
|
}()
|
|
|
|
if err := m.initialize(ctx, wait, handler); err != nil {
|
|
return err
|
|
}
|
|
|
|
waitCh := wait.start()
|
|
if waitCh != nil {
|
|
select {
|
|
case <-ctx.Done():
|
|
return context.Cause(ctx)
|
|
case err := <-monitorErr:
|
|
if err != nil {
|
|
return fmt.Errorf("watching file share sessions: %w", err)
|
|
} else if ctx.Err() == nil {
|
|
// this indicates a bug - it should not stop w/o an error if the context is still active
|
|
return errors.New("file share session watch stopped unexpectedly")
|
|
}
|
|
case <-wait.start():
|
|
// everything is done
|
|
}
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// initialize finds existing shares or creates new ones for the host paths.
|
|
//
|
|
// Once a share is found/created, its progress is monitored via the watch.
|
|
func (m *FileShareManager) initialize(ctx context.Context, wait *waiter, handler func(FileShareSession)) error {
|
|
// the watch is already running in the background, so the lock is taken
|
|
// throughout to prevent interleaving writes
|
|
m.mu.Lock()
|
|
defer m.mu.Unlock()
|
|
|
|
existing, err := m.cli.ListFileShares(ctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
for _, path := range m.hostPaths {
|
|
var fileShareID string
|
|
var fss *FileShareSession
|
|
|
|
if fss = findExistingShare(path, existing); fss != nil {
|
|
fileShareID = fss.Beta.Path
|
|
logrus.Debugf("Found existing suitable file share %s for path %q [%s]", fileShareID, path, fss.Alpha.Path)
|
|
wait.addShare(fileShareID)
|
|
handler(*fss)
|
|
continue
|
|
} else {
|
|
req := CreateFileShareRequest{
|
|
HostPath: path,
|
|
Labels: map[string]string{
|
|
"com.docker.compose.project": m.projectName,
|
|
},
|
|
}
|
|
createResp, err := m.cli.CreateFileShare(ctx, req)
|
|
if err != nil {
|
|
return fmt.Errorf("creating file share: %w", err)
|
|
}
|
|
fileShareID = createResp.FileShareID
|
|
fss = m.state[fileShareID]
|
|
logrus.Debugf("Created file share %s for path %q", fileShareID, path)
|
|
}
|
|
wait.addShare(fileShareID)
|
|
if fss != nil {
|
|
handler(*fss)
|
|
}
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
func (m *FileShareManager) watch(ctx context.Context, handler func(FileShareSession)) error {
|
|
events, err := m.cli.StreamFileShares(ctx)
|
|
if err != nil {
|
|
return fmt.Errorf("streaming file shares: %w", err)
|
|
}
|
|
|
|
for {
|
|
select {
|
|
case <-ctx.Done():
|
|
return nil
|
|
case event := <-events:
|
|
if event.Error != nil {
|
|
return fmt.Errorf("reading file share events: %w", event.Error)
|
|
}
|
|
// closure for lock
|
|
func() {
|
|
m.mu.Lock()
|
|
defer m.mu.Unlock()
|
|
for _, fss := range event.Value {
|
|
handler(fss)
|
|
}
|
|
}()
|
|
}
|
|
}
|
|
}
|
|
|
|
// eventHandler updates internal state, keeps track of in-progress syncs, and
|
|
// prints relevant events to progress.
|
|
func (m *FileShareManager) eventHandler(w progress.Writer, wait *waiter) func(fss FileShareSession) {
|
|
return func(fss FileShareSession) {
|
|
fileShareID := fss.Beta.Path
|
|
|
|
shouldPrint := wait.isWatching(fileShareID)
|
|
forProject := fss.Labels[api.ProjectLabel] == m.projectName
|
|
|
|
if shouldPrint || forProject {
|
|
m.state[fileShareID] = &fss
|
|
}
|
|
|
|
var percent int
|
|
var current, total int64
|
|
if fss.Beta.StagingProgress != nil {
|
|
current = int64(fss.Beta.StagingProgress.TotalReceivedSize)
|
|
} else {
|
|
current = int64(fss.Beta.TotalFileSize)
|
|
}
|
|
total = int64(fss.Alpha.TotalFileSize)
|
|
if total != 0 {
|
|
percent = int(current * 100 / total)
|
|
}
|
|
|
|
var status progress.EventStatus
|
|
var text string
|
|
|
|
switch {
|
|
case strings.HasPrefix(fss.Status, "halted"):
|
|
wait.shareDone(fileShareID)
|
|
status = progress.Error
|
|
case fss.Status == "watching":
|
|
wait.shareDone(fileShareID)
|
|
status = progress.Done
|
|
percent = 100
|
|
case fss.Status == "staging-beta":
|
|
status = progress.Working
|
|
// TODO(milas): the printer doesn't style statuses for children nicely
|
|
text = fmt.Sprintf(" Syncing (%7s / %-7s)",
|
|
units.HumanSize(float64(current)),
|
|
units.HumanSize(float64(total)),
|
|
)
|
|
default:
|
|
// catch-all for various other transitional statuses
|
|
status = progress.Working
|
|
}
|
|
|
|
evt := progress.Event{
|
|
ID: fss.Alpha.Path,
|
|
Status: status,
|
|
Text: text,
|
|
ParentID: fileShareProgressID,
|
|
Current: current,
|
|
Total: total,
|
|
Percent: percent,
|
|
}
|
|
|
|
if shouldPrint {
|
|
w.Event(evt)
|
|
}
|
|
}
|
|
}
|
|
|
|
func findExistingShare(path string, existing []FileShareSession) *FileShareSession {
|
|
for _, share := range existing {
|
|
if paths.IsChild(share.Alpha.Path, path) {
|
|
return &share
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
type waiter struct {
|
|
mu sync.Mutex
|
|
shareIDs map[string]struct{}
|
|
done chan struct{}
|
|
}
|
|
|
|
func (w *waiter) addShare(fileShareID string) {
|
|
w.mu.Lock()
|
|
defer w.mu.Unlock()
|
|
w.shareIDs[fileShareID] = struct{}{}
|
|
}
|
|
|
|
func (w *waiter) isWatching(fileShareID string) bool {
|
|
w.mu.Lock()
|
|
defer w.mu.Unlock()
|
|
_, ok := w.shareIDs[fileShareID]
|
|
return ok
|
|
}
|
|
|
|
// start returns a channel to wait for any outstanding shares to be ready.
|
|
//
|
|
// If no shares are registered when this is called, nil is returned.
|
|
func (w *waiter) start() <-chan struct{} {
|
|
w.mu.Lock()
|
|
defer w.mu.Unlock()
|
|
if len(w.shareIDs) == 0 {
|
|
return nil
|
|
}
|
|
if w.done == nil {
|
|
w.done = make(chan struct{})
|
|
}
|
|
return w.done
|
|
}
|
|
|
|
func (w *waiter) shareDone(fileShareID string) {
|
|
w.mu.Lock()
|
|
defer w.mu.Unlock()
|
|
|
|
delete(w.shareIDs, fileShareID)
|
|
if len(w.shareIDs) == 0 && w.done != nil {
|
|
close(w.done)
|
|
w.done = nil
|
|
}
|
|
}
|