[
https://issues.apache.org/jira/browse/BEAM-7065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16826096#comment-16826096
]
Alexandre Thenorio commented on BEAM-7065:
------------------------------------------
Thank you very much Robert. I have forked the SDK and updated it to allow for
unlifted combines (It can be found here
https://github.com/ByteFlinger/beam/commit/c94cf9a9bbcdb41ce8b55e27b3959ab043c96455),
and built another image based on that (alethenorio/beam-go:unlifted_combines)
And it no longer throws the error however I see no debug prints. I also see a
few of these warnings on stackdriver logs though I am uncertain whether they
are relevant
{noformat}
exception: "java.util.concurrent.ExecutionException:
java.lang.RuntimeException: Error received from SDK harness for instruction
-68: execution plan for -56 not found at
java.util.concurrent.CompletableFuture.reportGet(CompletableFuture.java:357) at
java.util.concurrent.CompletableFuture.get(CompletableFuture.java:1895) at
org.apache.beam.sdk.util.MoreFutures.get(MoreFutures.java:57) at
org.apache.beam.runners.dataflow.worker.fn.control.BeamFnMapTaskExecutor$SingularProcessBundleProgressTracker.updateProgress(BeamFnMapTaskExecutor.java:347)
at
org.apache.beam.runners.dataflow.worker.fn.control.BeamFnMapTaskExecutor$SingularProcessBundleProgressTracker.periodicProgressUpdate(BeamFnMapTaskExecutor.java:334)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at
java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308) at
java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
at
java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745) Caused by: java.lang.RuntimeException:
Error received from SDK harness for instruction -68: execution plan for -56 not
found at
org.apache.beam.runners.fnexecution.control.FnApiControlClient$ResponseStreamObserver.onNext(FnApiControlClient.java:157)
at
org.apache.beam.runners.fnexecution.control.FnApiControlClient$ResponseStreamObserver.onNext(FnApiControlClient.java:140)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.stub.ServerCalls$StreamingServerCallHandler$StreamingServerCallListener.onMessage(ServerCalls.java:248)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.ForwardingServerCallListener.onMessage(ForwardingServerCallListener.java:33)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.Contexts$ContextualizedServerCallListener.onMessage(Contexts.java:76)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.internal.ServerCallImpl$ServerStreamListenerImpl.messagesAvailable(ServerCallImpl.java:263)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.internal.ServerImpl$JumpToApplicationThreadServerStreamListener$1MessagesAvailable.runInContext(ServerImpl.java:683)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.internal.ContextRunnable.run(ContextRunnable.java:37)
at
org.apache.beam.vendor.grpc.v1p13p1.io.grpc.internal.SerializingExecutor.run(SerializingExecutor.java:123)
... 3 more{noformat}
I then wondered whether it had to do with the fact I don't have a window and
added it and also removed pubsub out of the equation to make sure nothing else
was the issue
{code:java}
package main
import (
"context"
"encoding/json"
"flag"
"reflect"
"time"
"github.com/apache/beam/sdks/go/pkg/beam"
"github.com/apache/beam/sdks/go/pkg/beam/core/graph/mtime"
"github.com/apache/beam/sdks/go/pkg/beam/core/graph/window"
"github.com/apache/beam/sdks/go/pkg/beam/log"
"github.com/apache/beam/sdks/go/pkg/beam/transforms/stats"
"github.com/apache/beam/sdks/go/pkg/beam/x/beamx"
)
type Sensor struct {
Name string
Value int
Timestamp int64
}
var (
now = time.Now()
data = []Sensor{
{Name: "temperature", Value: 20, Timestamp: now.Unix()},
{Name: "temperature", Value: 22, Timestamp: now.Add(1 *
time.Second).Unix()},
{Name: "temperature", Value: 24, Timestamp: now.Add(2 *
time.Second).Unix()},
{Name: "temperature", Value: 26, Timestamp: now.Add(3 *
time.Second).Unix()},
{Name: "humidity", Value: 10, Timestamp: now.Unix()},
{Name: "humidity", Value: 12, Timestamp: now.Add(1 *
time.Second).Unix()},
{Name: "humidity", Value: 14, Timestamp: now.Add(2 *
time.Second).Unix()},
{Name: "humidity", Value: 16, Timestamp: now.Add(3 *
time.Second).Unix()},
}
)
func init() {
beam.RegisterType(reflect.TypeOf((*addTimestampFn)(nil)).Elem())
beam.RegisterType(reflect.TypeOf((*Sensor)(nil)).Elem())
}
func main() {
flag.Parse()
beam.Init()
ctx := context.Background()
p := beam.NewPipeline()
s := p.Root()
// Emits fixed sensor data
// Returns PCollection<[]byte>
col := beam.Create(s, Bytes(data...)...)
// Transforms incoming bytes from pubsub to a *Sensor
// Accepts PCollection<[]byte>
// Returns PCollection<*Sensor>
sensor := beam.ParDo(s, extractSensorData, col)
// Injects the tag timestamp as the timestamp to be used by the window
operation
// Accepts PCollection<*Sensor>
// Returns PCollection<*Sensor> with window timestamps
tsSensor := beam.ParDo(s, &addTimestampFn{Min: mtime.Now()}, sensor)
// Transforms incoming bytes from pubsub to a string,int key value
// where the key is the sensor name and the value is the sensor reading
// Accepts PCollection<*Sensor>
// Returns PCollection<KV<string,int>>
data := beam.ParDo(s, extractKV, tsSensor)
// Split into windows of 2 seconds every second
window := beam.WindowInto(s, window.NewSlidingWindows(time.Second,
2*time.Second), data)
// Calculate running average per sensor
//
// Accpets PCollection<KV<string,int>>
// Returns PCollection<KV<string,int>>
avg := stats.MeanPerKey(s, window)
// Prints the values for each window average
beam.ParDo0(s, printValue, avg)
if err := beamx.Run(context.Background(), p); err != nil {
log.Exitf(ctx, "Failed to execute job: %v", err)
}
}
func Bytes(data ...Sensor) []interface{} {
bytes := []interface{}{}
for _, d := range data {
b, err := json.Marshal(&d)
if err != nil {
log.Fatal(context.Background(), err)
}
bytes = append(bytes, b)
}
return bytes
}
type addTimestampFn struct {
Min beam.EventTime `json:"min"`
}
func (f *addTimestampFn) ProcessElement(t *Sensor) (beam.EventTime, *Sensor) {
timestamp := mtime.FromTime(time.Unix(t.Timestamp, 0))
return timestamp, t
}
func printValue(key string, val float64) {
log.Infof(context.Background(), "Key %s, Val: %f", key, val)
}
func extractSensorData(msg []byte) *Sensor {
ctx := context.Background()
data := &Sensor{}
if err := json.Unmarshal(msg, data); err != nil {
log.Fatal(ctx, err)
}
return data
}
func extractKV(data *Sensor) (string, int) {
return data.Name, data.Value
}
{code}
This worked like a charm and those warnings were not there (And I can see the
prints on stackdriver)
I then proceeded to add pubsub back to this code instead of emitting the tags
with beam.Create and once again it did not work and the warnings were back.
{code:java}
package main
import (
"context"
"encoding/json"
"flag"
"fmt"
"reflect"
"time"
"cloud.google.com/go/pubsub"
"github.com/apache/beam/sdks/go/pkg/beam"
"github.com/apache/beam/sdks/go/pkg/beam/core/graph/mtime"
"github.com/apache/beam/sdks/go/pkg/beam/core/graph/window"
"github.com/apache/beam/sdks/go/pkg/beam/io/pubsubio"
"github.com/apache/beam/sdks/go/pkg/beam/log"
"github.com/apache/beam/sdks/go/pkg/beam/options/gcpopts"
"github.com/apache/beam/sdks/go/pkg/beam/transforms/stats"
"github.com/apache/beam/sdks/go/pkg/beam/util/pubsubx"
"github.com/apache/beam/sdks/go/pkg/beam/x/beamx"
)
var (
input = flag.String("input", "iot-data", "Pubsub input topic.")
)
type Sensor struct {
Name string
Value int
Timestamp int64
}
var (
now = time.Now()
data = []Sensor{
{Name: "temperature", Value: 20, Timestamp: now.Unix()},
{Name: "temperature", Value: 22, Timestamp: now.Add(1 *
time.Second).Unix()},
{Name: "temperature", Value: 24, Timestamp: now.Add(2 *
time.Second).Unix()},
{Name: "temperature", Value: 26, Timestamp: now.Add(3 *
time.Second).Unix()},
{Name: "humidity", Value: 10, Timestamp: now.Unix()},
{Name: "humidity", Value: 12, Timestamp: now.Add(1 *
time.Second).Unix()},
{Name: "humidity", Value: 14, Timestamp: now.Add(2 *
time.Second).Unix()},
{Name: "humidity", Value: 16, Timestamp: now.Add(3 *
time.Second).Unix()},
}
)
func init() {
beam.RegisterType(reflect.TypeOf((*addTimestampFn)(nil)).Elem())
beam.RegisterType(reflect.TypeOf((*Sensor)(nil)).Elem())
}
func main() {
flag.Parse()
beam.Init()
ctx := context.Background()
project := gcpopts.GetProject(ctx)
log.Infof(ctx, "Publishing %v messages to: %v", len(data), *input)
defer pubsubx.CleanupTopic(ctx, project, *input)
sub, err := Publish(ctx, project, *input, data...)
if err != nil {
log.Fatal(ctx, err)
}
log.Infof(ctx, "Running streaming sensor data with subscription: %v",
sub.ID())
p := beam.NewPipeline()
s := p.Root()
// Reads sensor data from pubsub
// Returns PCollection<[]byte>
col := pubsubio.Read(s, project, *input,
&pubsubio.ReadOptions{Subscription: sub.ID()})
// // Emits fixed sensor data
// // Returns PCollection<[]byte>
// col := beam.Create(s, Bytes(data...)...)
// Transforms incoming bytes from pubsub to a *Sensor
// Accepts PCollection<[]byte>
// Returns PCollection<*Sensor>
sensor := beam.ParDo(s, extractSensorData, col)
// Injects the tag timestamp as the timestamp to be used by the window
operation
// Accepts PCollection<*Sensor>
// Returns PCollection<*Sensor> with window timestamps
tsSensor := beam.ParDo(s, &addTimestampFn{Min: mtime.Now()}, sensor)
// Transforms incoming bytes from pubsub to a string,int key value
// where the key is the sensor name and the value is the sensor reading
// Accepts PCollection<*Sensor>
// Returns PCollection<KV<string,int>>
data := beam.ParDo(s, extractKV, tsSensor)
// Split into windows of 2 seconds every second
window := beam.WindowInto(s, window.NewSlidingWindows(time.Second,
2*time.Second), data)
// Calculate running average per sensor
//
// Accpets PCollection<KV<string,int>>
// Returns PCollection<KV<string,int>>
avg := stats.MeanPerKey(s, window)
// Prints the values for each window average
beam.ParDo0(s, printValue, avg)
if err := beamx.Run(context.Background(), p); err != nil {
log.Exitf(ctx, "Failed to execute job: %v", err)
}
}
func Bytes(data ...Sensor) []interface{} {
bytes := []interface{}{}
for _, d := range data {
b, err := json.Marshal(&d)
if err != nil {
log.Fatal(context.Background(), err)
}
bytes = append(bytes, b)
}
return bytes
}
type addTimestampFn struct {
Min beam.EventTime `json:"min"`
}
func (f *addTimestampFn) ProcessElement(t *Sensor) (beam.EventTime, *Sensor) {
timestamp := mtime.FromTime(time.Unix(t.Timestamp, 0))
return timestamp, t
}
func printValue(key string, val float64) {
log.Infof(context.Background(), "Key %s, Val: %f", key, val)
}
func extractSensorData(msg []byte) *Sensor {
ctx := context.Background()
data := &Sensor{}
if err := json.Unmarshal(msg, data); err != nil {
log.Fatal(ctx, err)
}
return data
}
func extractKV(data *Sensor) (string, int) {
return data.Name, data.Value
}
func Publish(ctx context.Context, project, topic string, messages ...Sensor)
(*pubsub.Subscription, error) {
client, err := pubsub.NewClient(ctx, project)
if err != nil {
return nil, err
}
t, err := pubsubx.EnsureTopic(ctx, client, topic)
if err != nil {
return nil, err
}
sub, err := pubsubx.EnsureSubscription(ctx, client, topic,
fmt.Sprintf("%v.sub.%v", topic, time.Now().Unix()))
if err != nil {
return nil, err
}
for _, msg := range messages {
s := &Sensor{}
bytes, err := json.Marshal(s)
if err != nil {
return nil, fmt.Errorf("failed to unmarshal '%v': %v", msg, err)
}
m := &pubsub.Message{
Data: ([]byte)(bytes),
// Attributes: ??
}
id, err := t.Publish(ctx, m).Get(ctx)
if err != nil {
return nil, fmt.Errorf("failed to publish '%v': %v", msg, err)
}
log.Infof(ctx, "Published %v with id: %v", msg, id)
}
return sub, nil
}
{code}
Not certain what the issue with pubsub here is but I guess this specific issue
can be closed for now. Any more info you might have on the issue with pubsub is
welcome
> Unable to use unlifted combine functions
> ----------------------------------------
>
> Key: BEAM-7065
> URL: https://issues.apache.org/jira/browse/BEAM-7065
> Project: Beam
> Issue Type: Bug
> Components: sdk-go
> Affects Versions: Not applicable
> Environment: Google Cloud Dataflow
> Reporter: Alexandre Thenorio
> Priority: Major
>
> I have tried running a simple example to calculate a running average or sum
> using the `stats` package however it does not seems to work.
>
> Here's a reproducer
>
> {code:java}
> package main
> import (
> "context"
> "encoding/json"
> "flag"
> "fmt"
> "time"
> "cloud.google.com/go/pubsub"
> "github.com/apache/beam/sdks/go/pkg/beam"
> "github.com/apache/beam/sdks/go/pkg/beam/io/pubsubio"
> "github.com/apache/beam/sdks/go/pkg/beam/log"
> "github.com/apache/beam/sdks/go/pkg/beam/options/gcpopts"
> "github.com/apache/beam/sdks/go/pkg/beam/transforms/stats"
> "github.com/apache/beam/sdks/go/pkg/beam/util/pubsubx"
> "github.com/apache/beam/sdks/go/pkg/beam/x/beamx"
> "github.com/apache/beam/sdks/go/pkg/beam/x/debug"
> )
> var (
> input = flag.String("input", "iot-data", "Pubsub input topic.")
> )
> type sensor struct {
> name string
> value int
> }
> var (
> data = []sensor{
> {name: "temperature", value: 24},
> {name: "humidity", value: 10},
> {name: "temperature", value: 20},
> {name: "temperature", value: 22},
> {name: "humidity", value: 14},
> {name: "humidity", value: 18},
> }
> )
> func main() {
> flag.Parse()
> beam.Init()
> ctx := context.Background()
> project := gcpopts.GetProject(ctx)
> log.Infof(ctx, "Publishing %v messages to: %v", len(data), *input)
> defer pubsubx.CleanupTopic(ctx, project, *input)
> sub, err := Publish(ctx, project, *input, data...)
> if err != nil {
> log.Fatal(ctx, err)
> }
> log.Infof(ctx, "Running streaming sensor data with subscription: %v",
> sub.ID())
> p := beam.NewPipeline()
> s := p.Root()
> // Reads sensor data from pubsub
> // Returns PCollection<[]byte>
> col := pubsubio.Read(s, project, *input,
> &pubsubio.ReadOptions{Subscription: sub.ID()})
> // Transforms incoming bytes from pubsub to a string,int key value
> // where the key is the sensor name and the value is the sensor reading
> // Accepts PCollection<[]byte>
> // Returns PCollection<KV<string,int>>
> data := beam.ParDo(s, extractSensorData, col)
> // Calculate running average per sensor
> //
> // Accpets PCollection<KV<string,int>>
> // Returns PCollection<KV<string,int>>
> sum := stats.MeanPerKey(s, data)
> debug.Print(s, sum)
> if err := beamx.Run(context.Background(), p); err != nil {
> log.Exitf(ctx, "Failed to execute job: %v", err)
> }
> }
> func extractSensorData(msg []byte) (string, int) {
> ctx := context.Background()
> data := &sensor{}
> if err := json.Unmarshal(msg, data); err != nil {
> log.Fatal(ctx, err)
> }
> return data.name, data.value
> }
> func Publish(ctx context.Context, project, topic string, messages ...sensor)
> (*pubsub.Subscription, error) {
> client, err := pubsub.NewClient(ctx, project)
> if err != nil {
> return nil, err
> }
> t, err := pubsubx.EnsureTopic(ctx, client, topic)
> if err != nil {
> return nil, err
> }
> sub, err := pubsubx.EnsureSubscription(ctx, client, topic,
> fmt.Sprintf("%v.sub.%v", topic, time.Now().Unix()))
> if err != nil {
> return nil, err
> }
> for _, msg := range messages {
> s := &sensor{}
> bytes, err := json.Marshal(s)
> if err != nil {
> return nil, fmt.Errorf("failed to unmarshal '%v': %v", msg, err)
> }
> m := &pubsub.Message{
> Data: ([]byte)(bytes),
> // Attributes: ??
> }
> id, err := t.Publish(ctx, m).Get(ctx)
> if err != nil {
> return nil, fmt.Errorf("failed to publish '%v': %v", msg, err)
> }
> log.Infof(ctx, "Published %v with id: %v", msg, id)
> }
> return sub, nil
> }
> {code}
>
> I ran this code in the following way
>
> {noformat}
> go run . --project="<my-project>" --runner dataflow --staging_location
> gs://<my-gs-bucket>/binaries/ --temp_location gs://<my-gs-bucket>/tmp/
> --region "europe-west1"
> --worker_harness_container_image=alethenorio/beam-go:v2.11.0{noformat}
>
>
>
> The code published to pubsub and then reads the messages and attempts to call
> `stats.MeanPerKey` to create a running average.
>
> When deploying this on cloud dataflow, using a container I built myself from
> the v2.11.0 version (alethenorio/beam-go:v2.11.0) I get the following error
> every time
>
>
> {noformat}
> Worker panic: Unexpected transform URN:
> beam:transform:combine_grouped_values:v1goroutine 1 [running]:
> runtime/debug.Stack(0x50, 0x0, 0x0)
> /usr/local/go/src/runtime/debug/stack.go:24 +0x9d
> runtime/debug.PrintStack()
> /usr/local/go/src/runtime/debug/stack.go:16 +0x22
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/harness/init.hook.func1()
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/harness/init/init.go:77
> +0xac
> panic(0xd4b160, 0xc001172f70)
> /usr/local/go/src/runtime/panic.go:522 +0x1b5
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/exec.(*builder).makeLink(0xc00078b980,
> 0xc0007aa2c0, 0x18, 0xc0007aa180, 0x17, 0x0, 0x400000000000040,
> 0xffffffffffffffff, 0x0, 0x0)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/exec/translate.go:521
> +0x308f
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/exec.(*builder).makePCollection(0xc00078b980,
> 0xc0007aa2c0, 0x18, 0xc0011775a0, 0xf, 0x0, 0x0)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/exec/translate.go:281
> +0x5ff
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/exec.UnmarshalPlan(0xc00078b2c0,
> 0xc00031a0f0, 0xd49820, 0xff5e10)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/exec/translate.go:71
> +0x393
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/harness.(*control).handleInstruction(0xc000304300,
> 0x1026be0, 0xc00031a0f0, 0xc000774880, 0xc00031a0f0)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/harness/harness.go:155
> +0x1ae
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/harness.Main.func2(0x1026be0,
> 0xc00031a0f0, 0xc000774880)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/harness/harness.go:114
> +0x1cf
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/harness.Main(0x1026be0,
> 0xc00031a0f0, 0x7ffe0437db7c, 0xf, 0x7ffe0437db9f, 0xf, 0x0, 0x0)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/harness/harness.go:129
> +0x786
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime/harness/init.hook()
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/harness/init/init.go:86
> +0xee
> github.com/apache/beam/sdks/go/pkg/beam/core/runtime.Init()
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/core/runtime/init.go:42
> +0x50
> github.com/apache/beam/sdks/go/pkg/beam.Init(...)
> /home/localuser/go/pkg/mod/github.com/apache/[email protected]+incompatible/sdks/go/pkg/beam/forward.go:111
> main.main()
> /home/localuser/reproducer/main.go:43 +0x8f
> {noformat}
>
> I realize the Go SDK is not stable yet but I was uncertain where to go to
> post this issue in case the devs are not aware so I hope this is fine.
> I get the feeling there is some issue with the gRPC requests sending the
> wrong URN but I couldn't find where in the code the `v1goroutine` gets set (I
> think it needs to be just v1)
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
