databricks-cli/libs/sync/sync.go

190 lines
4.2 KiB
Go
Raw Normal View History

package sync
import (
"context"
"fmt"
"time"
Add directory tracking to sync (#425) ## Changes This change replaces usage of the `repofiles` package with the `filer` package to consolidate WSFS code paths. The `repofiles` package implemented the following behavior. If a file at `foo/bar.txt` was created and removed, the directory `foo` was kept around because we do not perform directory tracking. If subsequently, a file at `foo` was created, it resulted in an `fs.ErrExist` because it is impossible to overwrite a directory. It would then perform a recursive delete of the path if this happened and retry the file write. To make this use case work without resorting to a recursive delete on conflict, we need to implement directory tracking as part of sync. The approach in this commit is as follows: 1. Maintain set of directories needed for current set of files. Compare to previous set of files. This results in mkdir of added directories and rmdir of removed directories. 2. Creation of new directories should happen prior to writing files. Otherwise, many file writes may race to create the same parent directories, resulting in additional API calls. Removal of existing directories should happen after removing files. 3. Making new directories can be deduped across common prefixes where only the longest prefix is created recursively. 4. Removing existing directories must happen sequentially, starting with the longest prefix. 5. Removal of directories is a best effort. It fails only if the directory is not empty, and if this happens we know something placed a file or directory manually, outside of sync. ## Tests * Existing integration tests pass (modified where it used to assert directories weren't cleaned up) * New integration test to confirm the inability to remove a directory doesn't fail the sync run
2023-06-12 11:44:00 +00:00
"github.com/databricks/cli/libs/filer"
"github.com/databricks/cli/libs/git"
"github.com/databricks/cli/libs/log"
"github.com/databricks/databricks-sdk-go"
)
type SyncOptions struct {
LocalPath string
RemotePath string
Full bool
SnapshotBasePath string
PollInterval time.Duration
WorkspaceClient *databricks.WorkspaceClient
Host string
}
type Sync struct {
*SyncOptions
Add directory tracking to sync (#425) ## Changes This change replaces usage of the `repofiles` package with the `filer` package to consolidate WSFS code paths. The `repofiles` package implemented the following behavior. If a file at `foo/bar.txt` was created and removed, the directory `foo` was kept around because we do not perform directory tracking. If subsequently, a file at `foo` was created, it resulted in an `fs.ErrExist` because it is impossible to overwrite a directory. It would then perform a recursive delete of the path if this happened and retry the file write. To make this use case work without resorting to a recursive delete on conflict, we need to implement directory tracking as part of sync. The approach in this commit is as follows: 1. Maintain set of directories needed for current set of files. Compare to previous set of files. This results in mkdir of added directories and rmdir of removed directories. 2. Creation of new directories should happen prior to writing files. Otherwise, many file writes may race to create the same parent directories, resulting in additional API calls. Removal of existing directories should happen after removing files. 3. Making new directories can be deduped across common prefixes where only the longest prefix is created recursively. 4. Removing existing directories must happen sequentially, starting with the longest prefix. 5. Removal of directories is a best effort. It fails only if the directory is not empty, and if this happens we know something placed a file or directory manually, outside of sync. ## Tests * Existing integration tests pass (modified where it used to assert directories weren't cleaned up) * New integration test to confirm the inability to remove a directory doesn't fail the sync run
2023-06-12 11:44:00 +00:00
fileSet *git.FileSet
snapshot *Snapshot
filer filer.Filer
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
// Synchronization progress events are sent to this event notifier.
notifier EventNotifier
seq int
}
// New initializes and returns a new [Sync] instance.
func New(ctx context.Context, opts SyncOptions) (*Sync, error) {
fileSet, err := git.NewFileSet(opts.LocalPath)
if err != nil {
return nil, err
}
err = fileSet.EnsureValidGitIgnoreExists()
if err != nil {
return nil, err
}
// Verify that the remote path we're about to synchronize to is valid and allowed.
err = EnsureRemotePathIsUsable(ctx, opts.WorkspaceClient, opts.RemotePath)
if err != nil {
return nil, err
}
// TODO: The host may be late-initialized in certain Azure setups where we
// specify the workspace by its resource ID. tracked in: https://databricks.atlassian.net/browse/DECO-194
opts.Host = opts.WorkspaceClient.Config.Host
if opts.Host == "" {
return nil, fmt.Errorf("failed to resolve host for snapshot")
}
// For full sync, we start with an empty snapshot.
// For incremental sync, we try to load an existing snapshot to start from.
var snapshot *Snapshot
if opts.Full {
snapshot, err = newSnapshot(ctx, &opts)
if err != nil {
return nil, fmt.Errorf("unable to instantiate new sync snapshot: %w", err)
}
} else {
snapshot, err = loadOrNewSnapshot(ctx, &opts)
if err != nil {
return nil, fmt.Errorf("unable to load sync snapshot: %w", err)
}
}
Add directory tracking to sync (#425) ## Changes This change replaces usage of the `repofiles` package with the `filer` package to consolidate WSFS code paths. The `repofiles` package implemented the following behavior. If a file at `foo/bar.txt` was created and removed, the directory `foo` was kept around because we do not perform directory tracking. If subsequently, a file at `foo` was created, it resulted in an `fs.ErrExist` because it is impossible to overwrite a directory. It would then perform a recursive delete of the path if this happened and retry the file write. To make this use case work without resorting to a recursive delete on conflict, we need to implement directory tracking as part of sync. The approach in this commit is as follows: 1. Maintain set of directories needed for current set of files. Compare to previous set of files. This results in mkdir of added directories and rmdir of removed directories. 2. Creation of new directories should happen prior to writing files. Otherwise, many file writes may race to create the same parent directories, resulting in additional API calls. Removal of existing directories should happen after removing files. 3. Making new directories can be deduped across common prefixes where only the longest prefix is created recursively. 4. Removing existing directories must happen sequentially, starting with the longest prefix. 5. Removal of directories is a best effort. It fails only if the directory is not empty, and if this happens we know something placed a file or directory manually, outside of sync. ## Tests * Existing integration tests pass (modified where it used to assert directories weren't cleaned up) * New integration test to confirm the inability to remove a directory doesn't fail the sync run
2023-06-12 11:44:00 +00:00
filer, err := filer.NewWorkspaceFilesClient(opts.WorkspaceClient, opts.RemotePath)
if err != nil {
return nil, err
}
return &Sync{
SyncOptions: &opts,
Add directory tracking to sync (#425) ## Changes This change replaces usage of the `repofiles` package with the `filer` package to consolidate WSFS code paths. The `repofiles` package implemented the following behavior. If a file at `foo/bar.txt` was created and removed, the directory `foo` was kept around because we do not perform directory tracking. If subsequently, a file at `foo` was created, it resulted in an `fs.ErrExist` because it is impossible to overwrite a directory. It would then perform a recursive delete of the path if this happened and retry the file write. To make this use case work without resorting to a recursive delete on conflict, we need to implement directory tracking as part of sync. The approach in this commit is as follows: 1. Maintain set of directories needed for current set of files. Compare to previous set of files. This results in mkdir of added directories and rmdir of removed directories. 2. Creation of new directories should happen prior to writing files. Otherwise, many file writes may race to create the same parent directories, resulting in additional API calls. Removal of existing directories should happen after removing files. 3. Making new directories can be deduped across common prefixes where only the longest prefix is created recursively. 4. Removing existing directories must happen sequentially, starting with the longest prefix. 5. Removal of directories is a best effort. It fails only if the directory is not empty, and if this happens we know something placed a file or directory manually, outside of sync. ## Tests * Existing integration tests pass (modified where it used to assert directories weren't cleaned up) * New integration test to confirm the inability to remove a directory doesn't fail the sync run
2023-06-12 11:44:00 +00:00
fileSet: fileSet,
snapshot: snapshot,
filer: filer,
notifier: &NopNotifier{},
seq: 0,
}, nil
}
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
func (s *Sync) Events() <-chan Event {
ch := make(chan Event, MaxRequestsInFlight)
s.notifier = &ChannelNotifier{ch}
return ch
}
func (s *Sync) Close() {
if s.notifier == nil {
return
}
s.notifier.Close()
s.notifier = nil
}
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
func (s *Sync) notifyStart(ctx context.Context, d diff) {
// If this is not the initial iteration we can ignore no-ops.
if s.seq > 0 && d.IsEmpty() {
return
}
s.notifier.Notify(ctx, newEventStart(s.seq, d.put, d.delete))
}
func (s *Sync) notifyProgress(ctx context.Context, action EventAction, path string, progress float32) {
s.notifier.Notify(ctx, newEventProgress(s.seq, action, path, progress))
}
func (s *Sync) notifyComplete(ctx context.Context, d diff) {
// If this is not the initial iteration we can ignore no-ops.
if s.seq > 0 && d.IsEmpty() {
return
}
s.notifier.Notify(ctx, newEventComplete(s.seq, d.put, d.delete))
s.seq++
}
func (s *Sync) RunOnce(ctx context.Context) error {
// tradeoff: doing portable monitoring only due to macOS max descriptor manual ulimit setting requirement
// https://github.com/gorakhargosh/watchdog/blob/master/src/watchdog/observers/kqueue.py#L394-L418
all, err := s.fileSet.All()
if err != nil {
log.Errorf(ctx, "cannot list files: %s", err)
return err
}
change, err := s.snapshot.diff(ctx, all)
if err != nil {
return err
}
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
s.notifyStart(ctx, change)
if change.IsEmpty() {
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
s.notifyComplete(ctx, change)
return nil
}
err = s.applyDiff(ctx, change)
if err != nil {
return err
}
err = s.snapshot.Save(ctx)
if err != nil {
log.Errorf(ctx, "cannot store snapshot: %s", err)
return err
}
Add optional JSON output for sync command (#230) JSON output makes it easy to process synchronization progress information in downstream tools (e.g. the vscode extension). This changes introduces a `sync.Event` interface type for progress events as well as an `sync.EventNotifier` that lets the sync code pass along progress events to calling code. Example output in text mode (default, this uses the existing logger calls): ```text 2023/03/03 14:07:17 [INFO] Remote file sync location: /Repos/pieter.noordhuis@databricks.com/... 2023/03/03 14:07:18 [INFO] Initial Sync Complete 2023/03/03 14:07:22 [INFO] Action: PUT: foo 2023/03/03 14:07:23 [INFO] Uploaded foo 2023/03/03 14:07:23 [INFO] Complete 2023/03/03 14:07:25 [INFO] Action: DELETE: foo 2023/03/03 14:07:25 [INFO] Deleted foo 2023/03/03 14:07:25 [INFO] Complete ``` Example output in JSON mode: ```json {"timestamp":"2023-03-03T14:08:15.459439+01:00","seq":0,"type":"start"} {"timestamp":"2023-03-03T14:08:15.459461+01:00","seq":0,"type":"complete"} {"timestamp":"2023-03-03T14:08:18.459821+01:00","seq":1,"type":"start","put":["foo"]} {"timestamp":"2023-03-03T14:08:18.459867+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:19.418696+01:00","seq":1,"type":"progress","action":"put","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:19.421397+01:00","seq":1,"type":"complete","put":["foo"]} {"timestamp":"2023-03-03T14:08:22.459238+01:00","seq":2,"type":"start","delete":["foo"]} {"timestamp":"2023-03-03T14:08:22.459268+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":0} {"timestamp":"2023-03-03T14:08:22.686413+01:00","seq":2,"type":"progress","action":"delete","path":"foo","progress":1} {"timestamp":"2023-03-03T14:08:22.688989+01:00","seq":2,"type":"complete","delete":["foo"]} ``` --------- Co-authored-by: shreyas-goenka <88374338+shreyas-goenka@users.noreply.github.com>
2023-03-08 09:27:19 +00:00
s.notifyComplete(ctx, change)
return nil
}
func (s *Sync) DestroySnapshot(ctx context.Context) error {
return s.snapshot.Destroy(ctx)
}
func (s *Sync) SnapshotPath() string {
return s.snapshot.SnapshotPath
}
func (s *Sync) RunContinuous(ctx context.Context) error {
ticker := time.NewTicker(s.PollInterval)
defer ticker.Stop()
for {
select {
case <-ctx.Done():
return ctx.Err()
case <-ticker.C:
err := s.RunOnce(ctx)
if err != nil {
return err
}
}
}
}