chore(agent.buffer): Add experimental warnings and increased logging (#15638)

config/config.go

@@ -279,7 +279,12 @@ type AgentConfig struct {
 	// startup. Set to -1 for unlimited attempts.
 	ConfigURLRetryAttempts int `toml:"config_url_retry_attempts"`
 
-	BufferStrategy  string `toml:"buffer_strategy"`
+	// BufferStrategy is the metric buffer type to use for a given output plugin.
+	// Supported types currently are "memory" and "disk".
+	BufferStrategy string `toml:"buffer_strategy"`
+
+	// BufferDirectory is the directory to store buffer files for serialized
+	// to disk metrics when using the "disk" buffer strategy.
 	BufferDirectory string `toml:"buffer_directory"`
 }
 
@@ -1531,6 +1536,10 @@ func (c *Config) buildOutput(name string, tbl *ast.Table) (*models.OutputConfig,
 		return nil, c.firstErr()
 	}
 
+	if oc.BufferStrategy == "disk" {
+		log.Printf("W! Using disk buffer strategy for plugin outputs.%s, this is an experimental feature", name)
+	}
+
 	// Generate an ID for the plugin
 	oc.ID, err = generatePluginID("outputs."+name, tbl)
 	return oc, err

docs/CONFIGURATION.md

@@ -352,6 +352,12 @@ The agent table configures Telegraf and the defaults used across all plugins.
   By default, processors are run a second time after aggregators. Changing
   this setting to true will skip the second run of processors.
 
+- **buffer_strategy**:
+  The type of buffer to use for telegraf output plugins. Supported modes are
+  `memory`, the default and original buffer type, and `disk`, an experimental
+  disk-backed buffer which will serialize all metrics to disk as needed to
+  improve data durability and reduce the chance for data loss.
+
 ## Plugins
 
 Telegraf plugins are divided into 4 types: [inputs][], [outputs][],

models/buffer_disk.go

@@ -3,6 +3,7 @@ package models
 import (
 	"errors"
 	"fmt"
+	"log"
 	"os"
 	"path/filepath"
 	"sync"
@@ -141,6 +142,7 @@ func (b *DiskBuffer) Batch(batchSize int) []telegraf.Metric {
 		}
 		if err != nil {
 			// non-recoverable error in deserialization, abort
+			log.Printf("E! raw metric data: %v", data)
 			panic(err)
 		}
 		if _, ok := m.(telegraf.TrackingMetric); ok && readIndex < b.originalEnd {
@@ -171,6 +173,7 @@ func (b *DiskBuffer) Accept(batch []telegraf.Metric) {
 	} else {
 		err := b.file.TruncateFront(b.batchFirst + uint64(len(batch)))
 		if err != nil {
+			log.Printf("E! batch length: %d, batchFirst: %d, batchSize: %d", len(batch), b.batchFirst, b.batchSize)
 			panic(err)
 		}
 	}