[exporter/splunk_hec] Apply compression if it's enabled in the config (#22969)

The compression used to be enabled only if the payload size was greater than 1.5KB which significantly complicated the logic and made it hard to test. This change makes the compression unconditionally applied to the payload if it's enabled in the config. The benchmarking shows improvements in the throughput and CPU usage for large payloads and expected degradation for small payloads which is acceptable given that it's not a common case.

Author: dmitryax

.chloggen/splunk-hec-apply-compression.yaml

@@ -0,0 +1,25 @@
+# Use this changelog template to create an entry for release notes.
+# If your change doesn't affect end users, such as a test fix or a tooling change,
+# you should instead start your pull request title with [chore] or use the "Skip Changelog" label.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: enhancement
+
+# The name of the component, or a single word describing the area of concern, (e.g. filelogreceiver)
+component: exporter/splunk_hec
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Apply compression to Splunk HEC payload unconditionally if it's enabled in the config.
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [22969, 22018]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+subtext: |
+  The compression used to be enabled only if the payload size was greater than 1.5KB which significantly
+  complicated the logic and made it hard to test. This change makes the compression unconditionally applied to 
+  the payload if it's enabled in the config. The benchmarking shows improvements in the throughput and CPU usage for 
+  large payloads and expected degradation for small payloads which is acceptable given that it's not a common case.
+

exporter/splunkhecexporter/buffer.go

@@ -17,22 +17,24 @@ var (
 	errOverCapacity = errors.New("over capacity")
 )
 
-// Minimum number of bytes to compress. 1500 is the MTU of an ethernet frame.
-const minCompressionLen = 1500
-
 // bufferState encapsulates intermediate buffer state when pushing data
 type bufferState struct {
-	compressionAvailable bool
-	bufferMaxLen         uint
-	maxEventLength       uint
-	writer               io.Writer
-	buf                  *bytes.Buffer
-	jsonStream           *jsoniter.Stream
-	rawLength            int
+	maxEventLength uint
+	buf            buffer
+	jsonStream     *jsoniter.Stream
+	rawLength      int
+}
+
+type buffer interface {
+	io.Writer
+	io.Reader
+	io.Closer
+	Reset()
+	Len() int
 }
 
 func (b *bufferState) compressionEnabled() bool {
-	_, ok := b.writer.(*cancellableGzipWriter)
+	_, ok := b.buf.(*cancellableGzipWriter)
 	return ok
 }
 
@@ -42,9 +44,6 @@ func (b *bufferState) containsData() bool {
 
 func (b *bufferState) reset() {
 	b.buf.Reset()
-	if _, ok := b.writer.(*cancellableBytesWriter); !ok {
-		b.writer = &cancellableBytesWriter{innerWriter: b.buf, maxCapacity: b.bufferMaxLen}
-	}
 	b.rawLength = 0
 }
 
@@ -53,64 +52,23 @@ func (b *bufferState) Read(p []byte) (n int, err error) {
 }
 
 func (b *bufferState) Close() error {
-	if _, ok := b.writer.(*cancellableGzipWriter); ok {
-		return b.writer.(*cancellableGzipWriter).close()
-	}
-	return nil
+	return b.buf.Close()
 }
 
 // accept returns true if data is accepted by the buffer
 func (b *bufferState) accept(data []byte) (bool, error) {
 	if len(data)+b.rawLength > int(b.maxEventLength) {
 		return false, nil
 	}
-	_, err := b.writer.Write(data)
-	overCapacity := errors.Is(err, errOverCapacity)
-	bufLen := b.buf.Len()
-	if overCapacity {
-		bufLen += len(data)
-	}
-	if b.compressionAvailable && !b.compressionEnabled() && bufLen > minCompressionLen {
-		// switch over to a zip buffer.
-		tmpBuf := bytes.NewBuffer(make([]byte, 0, b.bufferMaxLen+bufCapPadding))
-		writer := gzip.NewWriter(tmpBuf)
-		writer.Reset(tmpBuf)
-		zipWriter := &cancellableGzipWriter{
-			innerBuffer: tmpBuf,
-			innerWriter: writer,
-			// 8 bytes required for the zip footer.
-			maxCapacity: b.bufferMaxLen - 8,
-		}
-
-		if b.bufferMaxLen == 0 {
-			zipWriter.maxCapacity = 0
-		}
-
-		// we write the bytes buffer into the zip buffer. Any error from this is I/O, and should stop the process.
-		if _, err2 := zipWriter.Write(b.buf.Bytes()); err2 != nil {
-			return false, err2
-		}
-		b.writer = zipWriter
-		b.buf = tmpBuf
-		// if the byte writer was over capacity, try to write the new entry in the zip writer:
-		if overCapacity {
-			if _, err2 := zipWriter.Write(data); err2 != nil {
-				overCapacity2 := errors.Is(err2, errOverCapacity)
-				if overCapacity2 {
-					return false, nil
-				}
-				return false, err2
-			}
-
-		}
+	_, err := b.buf.Write(data)
+	if err == nil {
 		b.rawLength += len(data)
 		return true, nil
 	}
-	if overCapacity {
+	if errors.Is(err, errOverCapacity) {
 		return false, nil
 	}
-	b.rawLength += len(data)
-	return true, err
+	return false, err
 }
 
 type cancellableBytesWriter struct {
@@ -128,6 +86,22 @@ func (c *cancellableBytesWriter) Write(b []byte) (int, error) {
 	return c.innerWriter.Write(b)
 }
 
+func (c *cancellableBytesWriter) Read(p []byte) (int, error) {
+	return c.innerWriter.Read(p)
+}
+
+func (c *cancellableBytesWriter) Reset() {
+	c.innerWriter.Reset()
+}
+
+func (c *cancellableBytesWriter) Close() error {
+	return nil
+}
+
+func (c *cancellableBytesWriter) Len() int {
+	return c.innerWriter.Len()
+}
+
 type cancellableGzipWriter struct {
 	innerBuffer *bytes.Buffer
 	innerWriter *gzip.Writer
@@ -168,10 +142,24 @@ func (c *cancellableGzipWriter) Write(b []byte) (int, error) {
 	return c.innerWriter.Write(b)
 }
 
-func (c *cancellableGzipWriter) close() error {
+func (c *cancellableGzipWriter) Read(p []byte) (int, error) {
+	return c.innerBuffer.Read(p)
+}
+
+func (c *cancellableGzipWriter) Reset() {
+	c.innerBuffer.Reset()
+	c.innerWriter.Reset(c.innerBuffer)
+	c.len = 0
+}
+
+func (c *cancellableGzipWriter) Close() error {
 	return c.innerWriter.Close()
 }
 
+func (c *cancellableGzipWriter) Len() int {
+	return c.innerBuffer.Len()
+}
+
 // bufferStatePool is a pool of bufferState objects.
 type bufferStatePool struct {
 	pool *sync.Pool
@@ -189,18 +177,28 @@ func (p bufferStatePool) put(bf *bufferState) {
 
 const initBufferCap = 512
 
-func newBufferStatePool(bufCap uint, compressionAvailable bool, maxEventLength uint) bufferStatePool {
+func newBufferStatePool(bufCap uint, compressionEnabled bool, maxEventLength uint) bufferStatePool {
 	return bufferStatePool{
 		&sync.Pool{
 			New: func() interface{} {
-				buf := bytes.NewBuffer(make([]byte, 0, initBufferCap))
+				innerBuffer := bytes.NewBuffer(make([]byte, 0, initBufferCap))
+				var buf buffer
+				if compressionEnabled {
+					buf = &cancellableGzipWriter{
+						innerBuffer: innerBuffer,
+						innerWriter: gzip.NewWriter(buf),
+						maxCapacity: bufCap,
+					}
+				} else {
+					buf = &cancellableBytesWriter{
+						innerWriter: innerBuffer,
+						maxCapacity: bufCap,
+					}
+				}
 				return &bufferState{
-					compressionAvailable: compressionAvailable,
-					writer:               &cancellableBytesWriter{innerWriter: buf, maxCapacity: bufCap},
-					buf:                  buf,
-					jsonStream:           jsoniter.NewStream(jsoniter.ConfigDefault, nil, initBufferCap),
-					bufferMaxLen:         bufCap,
-					maxEventLength:       maxEventLength,
+					buf:            buf,
+					jsonStream:     jsoniter.NewStream(jsoniter.ConfigDefault, nil, initBufferCap),
+					maxEventLength: maxEventLength,
 				}
 			},
 		},

exporter/splunkhecexporter/client.go

@@ -219,7 +219,7 @@ func (c *client) fillLogsBuffer(logs plog.Logs, bs *bufferState, is iterState) (
 					}
 					permanentErrors = append(permanentErrors, consumererror.NewPermanent(
 						fmt.Errorf("dropped log event: error: event size %d bytes larger than configured max"+
-							" content length %d bytes", len(b), bs.bufferMaxLen)))
+							" content length %d bytes", len(b), c.config.MaxContentLengthLogs)))
 					return iterState{i, j, k + 1, false}, permanentErrors
 				}
 			}
@@ -277,7 +277,7 @@ func (c *client) fillMetricsBuffer(metrics pmetric.Metrics, bs *bufferState, is
 					}
 					permanentErrors = append(permanentErrors, consumererror.NewPermanent(
 						fmt.Errorf("dropped metric event: error: event size %d bytes larger than configured max"+
-							" content length %d bytes", len(b), bs.bufferMaxLen)))
+							" content length %d bytes", len(b), c.config.MaxContentLengthMetrics)))
 					return iterState{i, j, k + 1, false}, permanentErrors
 				}
 			}
@@ -322,7 +322,7 @@ func (c *client) fillTracesBuffer(traces ptrace.Traces, bs *bufferState, is iter
 					}
 					permanentErrors = append(permanentErrors, consumererror.NewPermanent(
 						fmt.Errorf("dropped span event: error: event size %d bytes larger than configured max"+
-							" content length %d bytes", len(b), bs.bufferMaxLen)))
+							" content length %d bytes", len(b), c.config.MaxContentLengthTraces)))
 					return iterState{i, j, k + 1, false}, permanentErrors
 				}
 			}
@@ -381,7 +381,7 @@ func (c *client) pushTracesDataInBatches(ctx context.Context, td ptrace.Traces,
 }
 
 func (c *client) postEvents(ctx context.Context, bufState *bufferState, headers map[string]string) error {
-	if err := bufState.Close(); err != nil {
+	if err := bufState.buf.Close(); err != nil {
 		return err
 	}
 	return c.hecWorker.send(ctx, bufState, headers)