Handle flow updates (#3455)

This commit is contained in:
Viktor Liu
2025-03-07 13:56:00 +01:00
committed by GitHub
parent 3c3a454e61
commit 54be772ffd
4 changed files with 154 additions and 94 deletions
+60 -51
View File
@@ -4,8 +4,10 @@ import (
"context"
"crypto/tls"
"crypto/x509"
"errors"
"fmt"
"strings"
"sync"
"time"
"github.com/cenkalti/backoff/v4"
@@ -25,95 +27,99 @@ type GRPCClient struct {
realClient proto.FlowServiceClient
clientConn *grpc.ClientConn
stream proto.FlowService_EventsClient
streamMu sync.Mutex
}
func NewClient(ctx context.Context, addr, payload, signature string) (*GRPCClient, error) {
func NewClient(addr, payload, signature string, interval time.Duration) (*GRPCClient, error) {
var opts []grpc.DialOption
transportOption := grpc.WithTransportCredentials(insecure.NewCredentials())
if strings.Contains(addr, "443") {
certPool, err := x509.SystemCertPool()
if err != nil || certPool == nil {
log.Debugf("System cert pool not available; falling back to embedded cert, error: %v", err)
certPool = embeddedroots.Get()
}
transportOption = grpc.WithTransportCredentials(credentials.NewTLS(&tls.Config{
opts = append(opts, grpc.WithTransportCredentials(credentials.NewTLS(&tls.Config{
RootCAs: certPool,
}))
})))
} else {
opts = append(opts, grpc.WithTransportCredentials(insecure.NewCredentials()))
}
connCtx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
conn, err := grpc.DialContext(
connCtx,
addr,
transportOption,
opts = append(opts,
nbgrpc.WithCustomDialer(),
grpc.WithBlock(),
grpc.WithIdleTimeout(interval*2),
grpc.WithKeepaliveParams(keepalive.ClientParameters{
Time: 30 * time.Second,
Timeout: 10 * time.Second,
}),
withAuthToken(payload, signature),
grpc.WithDefaultServiceConfig(`{"healthCheckConfig": {"serviceName": ""}}`),
)
conn, err := grpc.NewClient(addr, opts...)
if err != nil {
return nil, fmt.Errorf("dialing with context: %s", err)
return nil, fmt.Errorf("creating new grpc client: %w", err)
}
client := &GRPCClient{
return &GRPCClient{
realClient: proto.NewFlowServiceClient(conn),
clientConn: conn,
}
return client, nil
}, nil
}
func (c *GRPCClient) Close() error {
c.streamMu.Lock()
defer c.streamMu.Unlock()
c.stream = nil
return c.clientConn.Close()
}
func (c *GRPCClient) Receive(ctx context.Context, msgHandler func(msg *proto.FlowEventAck) error) error {
backOff := defaultBackoff(ctx)
func (c *GRPCClient) Receive(ctx context.Context, interval time.Duration, msgHandler func(msg *proto.FlowEventAck) error) error {
backOff := defaultBackoff(ctx, interval)
operation := func() error {
connState := c.clientConn.GetState()
if connState == connectivity.Shutdown {
return backoff.Permanent(fmt.Errorf("connection to signal has been shut down"))
}
stream, err := c.realClient.Events(ctx, grpc.WaitForReady(true))
if err != nil {
return err
}
c.stream = stream
err = checkHeader(stream)
if err != nil {
return err
}
return c.receive(stream, msgHandler)
return c.establishStreamAndReceive(ctx, msgHandler)
}
err := backoff.Retry(operation, backOff)
if err != nil {
log.Errorf("exiting the flow receiver service connection retry loop due to the unrecoverable error: %v", err)
return err
if err := backoff.Retry(operation, backOff); err != nil {
return fmt.Errorf("receive failed permanently: %w", err)
}
return nil
}
func (c *GRPCClient) establishStreamAndReceive(ctx context.Context, msgHandler func(msg *proto.FlowEventAck) error) error {
if c.clientConn.GetState() == connectivity.Shutdown {
return backoff.Permanent(errors.New("connection to flow receiver has been shut down"))
}
stream, err := c.realClient.Events(ctx, grpc.WaitForReady(true))
if err != nil {
return fmt.Errorf("create event stream: %w", err)
}
if err = checkHeader(stream); err != nil {
return fmt.Errorf("check header: %w", err)
}
c.streamMu.Lock()
c.stream = stream
c.streamMu.Unlock()
return c.receive(stream, msgHandler)
}
func (c *GRPCClient) receive(stream proto.FlowService_EventsClient, msgHandler func(msg *proto.FlowEventAck) error) error {
for {
msg, err := stream.Recv()
if err != nil {
return err
return fmt.Errorf("receive from stream: %w", err)
}
if err := msgHandler(msg); err != nil {
return err
return fmt.Errorf("handle message: %w", err)
}
}
}
@@ -122,7 +128,7 @@ func checkHeader(stream proto.FlowService_EventsClient) error {
header, err := stream.Header()
if err != nil {
log.Errorf("waiting for flow receiver header: %s", err)
return err
return fmt.Errorf("wait for header: %w", err)
}
if len(header) == 0 {
@@ -132,26 +138,29 @@ func checkHeader(stream proto.FlowService_EventsClient) error {
return nil
}
func defaultBackoff(ctx context.Context) backoff.BackOff {
func defaultBackoff(ctx context.Context, interval time.Duration) backoff.BackOff {
return backoff.WithContext(&backoff.ExponentialBackOff{
InitialInterval: 800 * time.Millisecond,
RandomizationFactor: 1,
Multiplier: 1.7,
MaxInterval: 10 * time.Second,
MaxInterval: interval / 2,
MaxElapsedTime: 3 * 30 * 24 * time.Hour, // 3 months
Stop: backoff.Stop,
Clock: backoff.SystemClock,
}, ctx)
}
func (c *GRPCClient) Send(ctx context.Context, event *proto.FlowEvent) error {
if c.stream == nil {
return fmt.Errorf("stream not initialized")
func (c *GRPCClient) Send(event *proto.FlowEvent) error {
c.streamMu.Lock()
stream := c.stream
c.streamMu.Unlock()
if stream == nil {
return errors.New("stream not initialized")
}
err := c.stream.Send(event)
if err != nil {
return fmt.Errorf("sending flow event: %s", err)
if err := stream.Send(event); err != nil {
return fmt.Errorf("send flow event: %w", err)
}
return nil