Skip to content
This repository has been archived by the owner on Nov 24, 2023. It is now read-only.

worker: fix potential panic when restoring subtask from meta #305

Merged
merged 3 commits into from
Oct 16, 2019
Merged

worker: fix potential panic when restoring subtask from meta #305

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e14e0b9
worker: fix potential panic when restoring subtask from meta
csuzhangxc Oct 8, 2019
a8c2e6a
Update dm/worker/worker.go
csuzhangxc Oct 14, 2019
713b989
Merge branch 'master' into fix-worker-ctx-panic
csuzhangxc Oct 14, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
51 changes: 33 additions & 18 deletions dm/worker/worker.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,14 +71,29 @@ type Worker struct {
}

// NewWorker creates a new Worker
func NewWorker(cfg *Config) (*Worker, error) {
w := &Worker{
func NewWorker(cfg *Config) (w *Worker, err error) {
w = &Worker{
cfg: cfg,
relayHolder: NewRelayHolder(cfg),
tracer: tracing.InitTracerHub(cfg.Tracer),
subTaskHolder: newSubTaskHolder(),
l: log.With(zap.String("component", "worker controller")),
}
w.ctx, w.cancel = context.WithCancel(context.Background())

defer func(w2 *Worker) {
if err != nil { // when err != nil, `w` will become nil in this func, so we pass `w` in defer.
// release resources, NOTE: we need to refactor New/Init/Start/Close for components later.
w2.cancel()
w2.subTaskHolder.closeAllSubTasks()
if w2.meta != nil {
w2.meta.Close()
}
if w2.db != nil {
w2.db.Close()
}
Comment on lines +86 to +94
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we just use w2.Close(), and maybe need add some check in Close (close something when it is not nil)

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we can.
But I also think we should refactor the pattern of components' new/init/start/close later.
like new only creates fields but does not acquire (IO) resources and start no goroutines, then when new failed, no operation needs to do and let GC to remove the objects is enough.

I think we need to do some refactoring when developing the HA version of DM.

what's your opinion?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK，we can do this later

}
}(w)

// initial relay holder
purger, err := w.relayHolder.Init([]purger.PurgeInterceptor{
Expand Down Expand Up @@ -107,16 +122,18 @@ func NewWorker(cfg *Config) (*Worker, error) {
}

// open kv db
w.db, err = openDB(dbDir, defaultKVConfig)
metaDB, err := openDB(dbDir, defaultKVConfig)
if err != nil {
return nil, err
}
w.db = metaDB

// initial metadata
w.meta, err = NewMetadata(dbDir, w.db)
meta, err := NewMetadata(dbDir, w.db)
if err != nil {
return nil, err
}
w.meta = meta

InitConditionHub(w)

Expand All @@ -125,6 +142,18 @@ func NewWorker(cfg *Config) (*Worker, error) {
return nil, err
}

w.l.Info("initialized")

return w, nil
}

// Start starts working
func (w *Worker) Start() {
if w.closed.Get() == closedTrue {
w.l.Warn("already closed")
return
}

// start relay
w.relayHolder.Start()

Expand All @@ -141,20 +170,6 @@ func NewWorker(cfg *Config) (*Worker, error) {
w.tracer.Start()
}

w.ctx, w.cancel = context.WithCancel(context.Background())

w.l.Info("initialzed")

return w, nil
}

// Start starts working
func (w *Worker) Start() {
if w.closed.Get() == closedTrue {
w.l.Warn("already closed")
return
}

w.wg.Add(2)
defer w.wg.Done()

Expand Down
6 changes: 3 additions & 3 deletions dm/worker/worker_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ func (t *testServer) testWorkerHandleTask(c *C) {
w.handleTask()
}()

c.Assert(utils.WaitSomething(5, 10*time.Millisecond, func() bool {
c.Assert(utils.WaitSomething(10, 100*time.Millisecond, func() bool {
w.meta.Lock()
defer w.meta.Unlock()
return len(w.meta.logs) == 0
Expand Down Expand Up @@ -154,7 +154,7 @@ func (t *testServer) TestTaskAutoResume(c *C) {
defer s.Close()
c.Assert(s.Start(), IsNil)
}()
c.Assert(utils.WaitSomething(30, 10*time.Millisecond, func() bool {
c.Assert(utils.WaitSomething(10, 100*time.Millisecond, func() bool {
return !s.closed.Get()
}), IsTrue)

Expand All @@ -165,7 +165,7 @@ func (t *testServer) TestTaskAutoResume(c *C) {
c.Assert(err, IsNil)

// check task in paused state
c.Assert(utils.WaitSomething(10, 10*time.Millisecond, func() bool {
c.Assert(utils.WaitSomething(10, 100*time.Millisecond, func() bool {
for _, st := range s.worker.QueryStatus(taskName) {
if st.Name == taskName && st.Stage == pb.Stage_Paused {
return true
Expand Down