-
Notifications
You must be signed in to change notification settings - Fork 111
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix: avoid unwanted watcher creation and reduce being stuck with udf is restarted #999
Merged
Changes from all commits
Commits
Show all changes
12 commits
Select commit
Hold shift + click to select a range
2af8623
avoid watcher creation when lastUpdateTime is equal to previousFetchTime
yhl25 ac5cea3
udpate comment
yhl25 045ef27
chore: fix comment
vigith 3565752
fix error handling for reduce
yhl25 34f39ab
Merge branch 'kvw' of github.com:numaproj/numaflow into kvw
yhl25 5879d7d
Apply suggestions from code review
yhl25 006bf5e
review comments
yhl25 4ecfb07
Merge branch 'main' into kvw
yhl25 6785c8d
revert
yhl25 dd5a88c
Merge branch 'kvw' of github.com:numaproj/numaflow into kvw
yhl25 1764278
fix flaky TestRead
yhl25 75e0eb9
use ctx
yhl25 File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -20,7 +20,6 @@ import ( | |
"context" | ||
"fmt" | ||
"strconv" | ||
"sync" | ||
"time" | ||
|
||
reducepb "github.com/numaproj/numaflow-go/pkg/apis/proto/reduce/v1" | ||
|
@@ -75,111 +74,101 @@ func (u *GRPCBasedReduce) WaitUntilReady(ctx context.Context) error { | |
|
||
// ApplyReduce accepts a channel of isbMessages and returns the aggregated result | ||
func (u *GRPCBasedReduce) ApplyReduce(ctx context.Context, partitionID *partition.ID, messageStream <-chan *isb.ReadMessage) ([]*isb.WriteMessage, error) { | ||
datumCh := make(chan *reducepb.ReduceRequest) | ||
var wg sync.WaitGroup | ||
var result *reducepb.ReduceResponse | ||
var err error | ||
var ( | ||
result *reducepb.ReduceResponse | ||
err error | ||
errCh = make(chan error, 1) | ||
responseCh = make(chan *reducepb.ReduceResponse, 1) | ||
datumCh = make(chan *reducepb.ReduceRequest) | ||
) | ||
|
||
// pass key and window information inside the context | ||
mdMap := map[string]string{ | ||
shared.WinStartTime: strconv.FormatInt(partitionID.Start.UnixMilli(), 10), | ||
shared.WinEndTime: strconv.FormatInt(partitionID.End.UnixMilli(), 10), | ||
} | ||
|
||
ctx = metadata.NewOutgoingContext(ctx, metadata.New(mdMap)) | ||
ctx, cancel := context.WithCancel(ctx) | ||
defer cancel() | ||
|
||
grpcCtx := metadata.NewOutgoingContext(ctx, metadata.New(mdMap)) | ||
|
||
// There can be two error scenarios: | ||
// 1. The u.client.ReduceFn method returns an error before reading all the messages from the messageStream | ||
// 2. The u.client.ReduceFn method returns an error after reading all the messages from the messageStream | ||
|
||
// invoke the reduceFn method with datumCh channel | ||
wg.Add(1) | ||
go func() { | ||
defer wg.Done() | ||
// TODO handle this error here itself | ||
result, err = u.client.ReduceFn(ctx, datumCh) | ||
result, err = u.client.ReduceFn(grpcCtx, datumCh) | ||
if err != nil { | ||
errCh <- err | ||
} else { | ||
responseCh <- result | ||
} | ||
close(errCh) | ||
close(responseCh) | ||
}() | ||
|
||
readLoop: | ||
for { | ||
select { | ||
case msg, ok := <-messageStream: | ||
if msg != nil { | ||
// create datum from isbMessage and send it to datumCh channel for reduceFn | ||
go func() { | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. go func() {
defer close(datumCh)
for {
select {
if xxx {
return
}
xxxx
}
}
}() There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. done |
||
// after reading all the messages from the messageStream or if ctx was canceled close the datumCh channel | ||
defer close(datumCh) | ||
for { | ||
select { | ||
case msg, ok := <-messageStream: | ||
// if the messageStream is closed or if the message is nil, return | ||
if !ok || msg == nil { | ||
return | ||
} | ||
|
||
d := createDatum(msg) | ||
|
||
// send the datum to datumCh channel, handle the case when the context is canceled | ||
select { | ||
case datumCh <- d: | ||
case <-ctx.Done(): | ||
close(datumCh) | ||
return nil, ctx.Err() | ||
return | ||
} | ||
} | ||
if !ok { | ||
break readLoop | ||
} | ||
case <-ctx.Done(): | ||
close(datumCh) | ||
return nil, ctx.Err() | ||
} | ||
} | ||
|
||
// close the datumCh, let the reduceFn know that there are no more messages | ||
close(datumCh) | ||
|
||
wg.Wait() | ||
|
||
if err != nil { | ||
// if any error happens in reduce | ||
// will exit and restart the numa container | ||
udfErr, _ := sdkerr.FromError(err) | ||
switch udfErr.ErrorKind() { | ||
case sdkerr.Retryable: | ||
// TODO: currently we don't handle retryable errors for reduce | ||
return nil, ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
} | ||
case sdkerr.NonRetryable: | ||
return nil, ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
} | ||
default: | ||
return nil, ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
case <-ctx.Done(): // if the context is done, return | ||
return | ||
} | ||
} | ||
} | ||
}() | ||
|
||
taggedMessages := make([]*isb.WriteMessage, 0) | ||
for _, response := range result.GetResults() { | ||
keys := response.Keys | ||
taggedMessage := &isb.WriteMessage{ | ||
Message: isb.Message{ | ||
Header: isb.Header{ | ||
MessageInfo: isb.MessageInfo{ | ||
EventTime: partitionID.End.Add(-1 * time.Millisecond), | ||
IsLate: false, | ||
// wait for the reduceFn to finish | ||
for { | ||
select { | ||
case err = <-errCh: | ||
if err != nil { | ||
return nil, convertToUdfError(err) | ||
} | ||
case result = <-responseCh: | ||
taggedMessages := make([]*isb.WriteMessage, 0) | ||
for _, response := range result.GetResults() { | ||
keys := response.Keys | ||
taggedMessage := &isb.WriteMessage{ | ||
Message: isb.Message{ | ||
Header: isb.Header{ | ||
MessageInfo: isb.MessageInfo{ | ||
EventTime: partitionID.End.Add(-1 * time.Millisecond), | ||
IsLate: false, | ||
}, | ||
Keys: keys, | ||
}, | ||
Body: isb.Body{ | ||
Payload: response.Value, | ||
}, | ||
}, | ||
Keys: keys, | ||
}, | ||
Body: isb.Body{ | ||
Payload: response.Value, | ||
}, | ||
}, | ||
Tags: response.Tags, | ||
Tags: response.Tags, | ||
} | ||
taggedMessages = append(taggedMessages, taggedMessage) | ||
} | ||
return taggedMessages, nil | ||
case <-ctx.Done(): | ||
return nil, convertToUdfError(ctx.Err()) | ||
} | ||
taggedMessages = append(taggedMessages, taggedMessage) | ||
} | ||
return taggedMessages, nil | ||
} | ||
|
||
func createDatum(readMessage *isb.ReadMessage) *reducepb.ReduceRequest { | ||
|
@@ -194,3 +183,40 @@ func createDatum(readMessage *isb.ReadMessage) *reducepb.ReduceRequest { | |
} | ||
return d | ||
} | ||
|
||
// convertToUdfError converts the error returned by the reduceFn to ApplyUDFErr | ||
func convertToUdfError(err error) ApplyUDFErr { | ||
// if any error happens in reduce | ||
// will exit and restart the numa container | ||
udfErr, _ := sdkerr.FromError(err) | ||
switch udfErr.ErrorKind() { | ||
case sdkerr.Retryable: | ||
// TODO: currently we don't handle retryable errors for reduce | ||
return ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
} | ||
case sdkerr.NonRetryable: | ||
return ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
} | ||
default: | ||
return ApplyUDFErr{ | ||
UserUDFErr: false, | ||
Message: fmt.Sprintf("gRPC client.ReduceFn failed, %s", err), | ||
InternalErr: InternalErr{ | ||
Flag: true, | ||
MainCarDown: false, | ||
}, | ||
} | ||
} | ||
} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
no need of wait group?