hyperledger/fabric · error

stream %d aborted

Error message

stream %d aborted

What it means

Stream.SendWithReport checks whether the stream was canceled before sending; if so, it returns an error whose message is the stored abortReason, typically "stream N aborted". Streams are aborted when the remote node fails, the connection dies, or the stream is reaped due to inactivity/queue overflow, so consensus messages or submit forwards cannot be delivered.

Source

Thrown at orderer/common/cluster/stream.go:59

}

// StreamOperation denotes an operation done by a stream, such a Send or Receive.
type StreamOperation func() (*orderer.StepResponse, error)

// Canceled returns whether the stream was canceled.
func (stream *Stream) Canceled() bool {
	return atomic.LoadUint32(stream.canceled) == uint32(1)
}

// Send sends the given request to the remote cluster member.
func (stream *Stream) Send(request *orderer.StepRequest) error {
	return stream.SendWithReport(request, func(_ error) {})
}

// SendWithReport sends the given request to the remote cluster member and invokes report on the send result.
func (stream *Stream) SendWithReport(request *orderer.StepRequest, report func(error)) error {
	if stream.Canceled() {
		return errors.New(stream.abortReason.Load().(string))
	}
	var allowDrop bool
	// We want to drop consensus transactions if the remote node cannot keep up with us,
	// otherwise we'll slow down the entire FSM.
	if request.GetConsensusRequest() != nil {
		allowDrop = true
	}

	return stream.sendOrDrop(request, allowDrop, report)
}

// sendOrDrop sends the given request to the remote cluster member, or drops it
// if it is a consensus request and the queue is full.
func (stream *Stream) sendOrDrop(request *orderer.StepRequest, allowDrop bool, report func(error)) error {
	msgType := "transaction"
	if allowDrop {
		msgType = "consensus"
	}

View on GitHub (pinned to 2736b63f8f)

Solutions

1. Check connectivity and health of the remote ordering node; restart it if it crashed
2. Verify TLS certificates and mutual auth between cluster members
3. Increase SendBufferSize / address remote slowness so streams are not aborted for being unable to keep up
4. Retry the send — the stream layer will establish a new stream on the next connection to the recovered node

if stream.Canceled() { return fmt.Errorf("stream aborted: %s", stream.AbortReason()) }

err := stream.Send(consensusReq)
if err != nil {
    if strings.Contains(err.Error(), "aborted") {
        // stream dead: reconnect and resend on a fresh stream
        return reconnectAndResend(consensusReq)
    }
    return err
}

When it happens

Trigger: Send is called on a Stream that was already canceled (stream.Canceled() true) because the remote endpoint aborted it — e.g. worker pool detected a dropped connection, the abort futures fired, or the connection was closed by the remote node.

Common situations: Remote ordering node crashed or restarted while the local node still held a stream to it; network partition; remote node too slow to drain consensus messages and its connection culled; TLS handshake failure mid-session aborting the stream.

Related errors

• response is of type %v, but expected a block
• orderer `%s` hung up without sending status
• invalid request object
• request message is nil
• stream %d is stale

AI-assisted analysis of hyperledger/fabric@2736b63f8f (2026-09-04). Data as JSON: /api/errors/e031a552d9375880. Report an issue: GitHub.