[exporter/loadbalancing] Add top level sending_queue, retry_on_failure and timeout settings (open-telemetry#36094)

#### Description
##### Problem statement
`loadbalancing` exporter is actually a wrapper that's creates and
manages set of actual `otlp` exporters
Those `otlp` exporters technically shares same configuration parameters
that are defined on `loadbalancing` exporter level, including
`sending_queue` configuration. The only difference is `endpoint`
parameter that are substituted by `loadbalancing` exporter itself
This means, that `sending_queue`, `retry_on_failure` and `timeout`
settings can be defined only on `otlp` sub-exporters, while top-level
`loadbalancing` exporter is missing all those settings
This configuration approach produces several issue, that are already
reported by users:
* Impossibility to use Persistent Queue in `loadbalancing` exporter (see
open-telemetry#16826). That's happens because `otlp` sub-exporters are sharing the
same configurations, including configuration of the queue, i.e. they all
are using the same `storage` instance at the same time which is not
possible at the moment
* Data loss even using `sending_queue` configuration (see open-telemetry#35378).
That's happens because Queue is defined on level of `otlp` sub-exporters
and if this exporter cannot flush data from queue (for example, endpoint
is not available anymore) there is no other options that just to discard
data from queue, i.e. there is no higher level queue and persistent
storage where data can be returned is case of permanent failure

There might be some other potential issue that was already tracked and
related to current configuration approach

##### Proposed solution
The easiest way to solve issues above - is to use standard approach for
queue, retry and timeout configuration using `exporterhelper`
This will bring queue, retry and timeout functionality to the top-level
of `loadbalancing` exporter, instead of `otlp` sub-exporters
Related to mentioned issues it will bring:
* Single Persistent Queue, that is used by all `otlp` sub-exporters (not
directly of course)
* Queue will not be discarded/destroyed if any (or all) of endpoint that
are unreachable anymore, top-level queue will keep data until new
endpoints will be available
* Scale-up and scale-down event for next layer of OpenTelemetry
Collectors in K8s environments will be more predictable, and will not
include data loss anymore (potential fix for open-telemetry#33959). There is still a
big chance of inconsistency when some data will be send to incorrect
endpoint, but it's already better state that we have right now

##### Noticeable changes
* `loadbalancing` exporter on top-level now uses `exporterhelper` with
all supported functionality by it
* `sending_queue` will be automatically disabled on `otlp` exporters
when it already present on top-level `loadbalancing` exporter. This
change is done to prevent data loss on `otlp` exporters because queue
there doesn't provide expected result. Also it will prevent potential
misconfiguration from user side and as result - irrelevant reported
issues
* `exporter` attribute for metrics generated from `otlp` sub-exporters
now includes endpoint for better visibility and to segregate them from
top-level `loadbalancing` exporter - was `"exporter": "loadbalancing"`,
now `"exporter": "loadbalancing/127.0.0.1:4317"`
* logs, generated by `otlp` sub-exporters now includes additional
attribute `endpoint` with endpoint value with the same reasons as for
metrics

#### Link to tracking issue
Fixes open-telemetry#35378
Fixes open-telemetry#16826

#### Testing
Proposed changes was heavily tested on large K8s environment with set of
different scale-up/scale-down event using persistent queue configuration
- no data loss were detected, everything works as expected

#### Documentation
`README.md` was updated to reflect new configuration parameters
available. Sample `config.yaml` was updated as well

Author: Fiery-Fenix

.chloggen/feat_loadbalancing-exporter-queue.yaml

@@ -0,0 +1,29 @@
+# Use this changelog template to create an entry for release notes.
+
+# 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: loadbalancingexporter
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Adding sending_queue, retry_on_failure and timeout settings to loadbalancing exporter configuration
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [35378,16826]
+
+# (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: |
+  When switching to top-level sending_queue configuration - users should carefully review queue size
+  In some rare cases setting top-level queue size to n*queueSize might be not enough to prevent data loss
+
+# If your change doesn't affect end users or the exported elements of any package,
+# you should instead start your pull request title with [chore] or use the "Skip Changelog" label.
+# Optional: The change log or logs in which this entry should be included.
+# e.g. '[user]' or '[user, api]'
+# Include 'user' if the change is relevant to end users.
+# Include 'api' if there is a change to a library API.
+# Default: '[user]'
+change_logs: [user]

exporter/loadbalancingexporter/README.md

@@ -48,14 +48,39 @@ This also supports service name based exporting for traces. If you have two or m
 
 ## Resilience and scaling considerations
 
-The `loadbalancingexporter` will, irrespective of the chosen resolver (`static`, `dns`, `k8s`), create one exporter per endpoint. The exporter conforms to its published configuration regarding sending queue and retry mechanisms. Importantly, the `loadbalancingexporter` will not attempt to re-route data to a healthy endpoint on delivery failure, and data loss is therefore possible if the exporter's target remains unavailable once redelivery is exhausted. Due consideration needs to be given to the exporter queue and retry configuration when running in a highly elastic environment.
+The `loadbalancingexporter` will, irrespective of the chosen resolver (`static`, `dns`, `k8s`), create one `otlp` exporter per endpoint. Each level of exporters, `loadbalancingexporter` itself and all sub-exporters (one per each endpoint), have it's own queue, timeout and retry mechanisms. Importantly, the `loadbalancingexporter`, by default, will NOT attempt to re-route data to a healthy endpoint on delivery failure, because in-memory queue, retry and timeout setting are disabled by default ([more details on queuing, retry and timeout default settings](https://github.com/open-telemetry/opentelemetry-collector/blob/main/exporter/exporterhelper/README.md)).
 
-* When using the `static` resolver and a target is unavailable, all the target's load-balanced telemetry will fail to be delivered until either the target is restored or removed from the static list. The same principle applies to the `dns` resolver.
+```
+                                        +------------------+          +---------------+
+ resiliency options 1                   |                  |          |               |
+                                       -- otlp exporter 1  ------------  backend 1    |
+           |                       ---/ |                  |          |               |
+           |                   ---/     +----|-------------+          +---------------+
+           |               ---/              |
+  +-----------------+  ---/                  |
+  |                 --/                      |
+  |  loadbalancing  |                   resiliency options 2
+  |    exporter     |                        |
+  |                 --\                      |
+  +-----------------+  ----\                 |
+                            ----\       +----|-------------+          +---------------+
+                                 ----\  |                  |          |               |
+                                      --- otlp exporter N  ------------  backend N    |
+                                        |                  |          |               |
+                                        +------------------+          +---------------+
+```
+
+* For all types of resolvers (`static`, `dns`, `k8s`) - if one of endpoints is unavailable - first works queue, retry and timeout settings defined for sub-exporters (under `otlp` property). Once redelivery is exhausted on sub-exporter level, and resilience options 1 are enabled - telemetry data returns to `loadbalancingexporter` itself and data redelivery happens according to exporter level queue, retry and timeout settings.
+* When using the `static` resolver and all targets are unavailable, all load-balanced telemetry will fail to be delivered until either one or all targets are restored or valid target is added the static list. The same principle applies to the `dns` and `k8s` resolvers, except for endpoints list update which happens automatically.
 * When using `k8s`, `dns`, and likely future resolvers, topology changes are eventually reflected in the `loadbalancingexporter`. The `k8s` resolver will update more quickly than `dns`, but a window of time in which the true topology doesn't match the view of the `loadbalancingexporter` remains.
+* Resiliency options 1 (`timeout`, `retry_on_failure` and `sending_queue` settings in `loadbalancing` section) - are useful for highly elastic environment (like k8s), where list of resolved endpoints frequently changed due to deployments, scale-up or scale-down events. In case of permanent change of list of resolved exporters this options provide capability to re-route data into new set of healthy backends. Disabled by default.
+* Resiliency options 1 (`timeout`, `retry_on_failure` and `sending_queue` settings in `otlp` section) - are useful for temporary problems with specific backend, like network flukes. Persistent Queue is NOT supported here as all sub-exporter shares the same `sending_queue` configuration, including `storage`. Enabled by default.
+
+Unfortunately, data loss is still possible if all of the exporter's targets remains unavailable once redelivery is exhausted. Due consideration needs to be given to the exporter queue and retry configuration when running in a highly elastic environment.
 
 ## Configuration
 
-Refer to [config.yaml](./testdata/config.yaml) for detailed examples on using the processor.
+Refer to [config.yaml](./testdata/config.yaml) for detailed examples on using the exporter.
 
 * The `otlp` property configures the template used for building the OTLP exporter. Refer to the OTLP Exporter documentation for information on which options are available. Note that the `endpoint` property should not be set and will be overridden by this exporter with the backend endpoint.
 * The `resolver` accepts a `static` node, a `dns`, a `k8s` service or `aws_cloud_map`. If all four are specified, an `errMultipleResolversProvided` error will be thrown.
@@ -90,6 +115,7 @@ Refer to [config.yaml](./testdata/config.yaml) for detailed examples on using th
   * `traceID`: Routes spans based on their `traceID`. Invalid for metrics.
   * `metric`: Routes metrics based on their metric name. Invalid for spans.
   * `streamID`: Routes metrics based on their datapoint streamID. That's the unique hash of all it's attributes, plus the attributes and identifying information of its resource, scope, and metric data
+* loadbalancing exporter supports set of standard [queuing, retry and timeout settings](https://github.com/open-telemetry/opentelemetry-collector/blob/main/exporter/exporterhelper/README.md), but they are disable by default to maintain compatibility
 
 Simple example
 
@@ -117,11 +143,76 @@ exporters:
         - backend-2:4317
         - backend-3:4317
         - backend-4:4317
-      # Notice to config a headless service DNS in Kubernetes  
+      # Notice to config a headless service DNS in Kubernetes
+      # dns:
+      #  hostname: otelcol-headless.observability.svc.cluster.local
+
+service:
+  pipelines:
+    traces:
+      receivers:
+        - otlp
+      processors: []
+      exporters:
+        - loadbalancing
+    logs:
+      receivers:
+        - otlp
+      processors: []
+      exporters:
+        - loadbalancing
+```
+
+Persistent queue, retry and timeout usage example:
+
+```yaml
+receivers:
+  otlp:
+    protocols:
+      grpc:
+        endpoint: localhost:4317
+
+processors:
+
+exporters:
+  loadbalancing:
+    timeout: 10s
+    retry_on_failure:
+      enabled: true
+      initial_interval: 5s
+      max_interval: 30s
+      max_elapsed_time: 300s
+    sending_queue:
+      enabled: true
+      num_consumers: 2
+      queue_size: 1000
+      storage: file_storage/otc
+    routing_key: "service"
+    protocol:
+      otlp:
+        # all options from the OTLP exporter are supported
+        # except the endpoint
+        timeout: 1s
+        sending_queue:
+          enabled: true
+    resolver:
+      static:
+        hostnames:
+        - backend-1:4317
+        - backend-2:4317
+        - backend-3:4317
+        - backend-4:4317
+      # Notice to config a headless service DNS in Kubernetes
       # dns:
-      #  hostname: otelcol-headless.observability.svc.cluster.local        
+      #  hostname: otelcol-headless.observability.svc.cluster.local
+
+extensions:
+  file_storage/otc:
+    directory: /var/lib/storage/otc
+    timeout: 10s
 
 service:
+  extensions: [file_storage]
   pipelines:
     traces:
       receivers:
@@ -334,7 +425,7 @@ service:
 
 ## Metrics
 
-The following metrics are recorded by this processor:
+The following metrics are recorded by this exporter:
 
 * `otelcol_loadbalancer_num_resolutions` represents the total number of resolutions performed by the resolver specified in the tag `resolver`, split by their outcome (`success=true|false`). For the static resolver, this should always be `1` with the tag `success=true`.
 * `otelcol_loadbalancer_num_backends` informs how many backends are currently in use. It should always match the number of items specified in the configuration file in case the `static` resolver is used, and should eventually (seconds) catch up with the DNS changes. Note that DNS caches that might exist between the load balancer and the record authority will influence how long it takes for the load balancer to see the change.

exporter/loadbalancingexporter/config.go

@@ -7,6 +7,8 @@ import (
 	"time"
 
 	"github.com/aws/aws-sdk-go-v2/service/servicediscovery/types"
+	"go.opentelemetry.io/collector/config/configretry"
+	"go.opentelemetry.io/collector/exporter/exporterhelper"
 	"go.opentelemetry.io/collector/exporter/otlpexporter"
 )
 
@@ -30,6 +32,10 @@ const (
 
 // Config defines configuration for the exporter.
 type Config struct {
+	TimeoutSettings           exporterhelper.TimeoutConfig `mapstructure:",squash"`
+	configretry.BackOffConfig `mapstructure:"retry_on_failure"`
+	QueueSettings             exporterhelper.QueueConfig `mapstructure:"sending_queue"`
+
 	Protocol   Protocol         `mapstructure:"protocol"`
 	Resolver   ResolverSettings `mapstructure:"resolver"`
 	RoutingKey string           `mapstructure:"routing_key"`

exporter/loadbalancingexporter/factory.go

@@ -7,14 +7,21 @@ package loadbalancingexporter // import "github.com/open-telemetry/opentelemetry
 
 import (
 	"context"
+	"fmt"
 
 	"go.opentelemetry.io/collector/component"
 	"go.opentelemetry.io/collector/exporter"
+	"go.opentelemetry.io/collector/exporter/exporterhelper"
 	"go.opentelemetry.io/collector/exporter/otlpexporter"
+	"go.uber.org/zap"
 
 	"github.com/open-telemetry/opentelemetry-collector-contrib/exporter/loadbalancingexporter/internal/metadata"
 )
 
+const (
+	zapEndpointKey = "endpoint"
+)
+
 // NewFactory creates a factory for the exporter.
 func NewFactory() exporter.Factory {
 	return exporter.NewFactory(
@@ -32,20 +39,110 @@ func createDefaultConfig() component.Config {
 	otlpDefaultCfg.Endpoint = "placeholder:4317"
 
 	return &Config{
+		// By default we disable resilience options on loadbalancing exporter level
+		// to maintain compatibility with workflow in previous versions
 		Protocol: Protocol{
 			OTLP: *otlpDefaultCfg,
 		},
 	}
 }
 
-func createTracesExporter(_ context.Context, params exporter.Settings, cfg component.Config) (exporter.Traces, error) {
-	return newTracesExporter(params, cfg)
+func buildExporterConfig(cfg *Config, endpoint string) otlpexporter.Config {
+	oCfg := cfg.Protocol.OTLP
+	oCfg.Endpoint = endpoint
+
+	return oCfg
+}
+
+func buildExporterSettings(params exporter.Settings, endpoint string) exporter.Settings {
+	// Override child exporter ID to segregate metrics from loadbalancing top level
+	childName := endpoint
+	if params.ID.Name() != "" {
+		childName = fmt.Sprintf("%s_%s", params.ID.Name(), childName)
+	}
+	params.ID = component.NewIDWithName(params.ID.Type(), childName)
+	// Add "endpoint" attribute to child exporter logger to segregate logs from loadbalancing top level
+	params.Logger = params.Logger.With(zap.String(zapEndpointKey, endpoint))
+
+	return params
+}
+
+func buildExporterResilienceOptions(options []exporterhelper.Option, cfg *Config) []exporterhelper.Option {
+	if cfg.TimeoutSettings.Timeout > 0 {
+		options = append(options, exporterhelper.WithTimeout(cfg.TimeoutSettings))
+	}
+	if cfg.QueueSettings.Enabled {
+		options = append(options, exporterhelper.WithQueue(cfg.QueueSettings))
+	}
+	if cfg.BackOffConfig.Enabled {
+		options = append(options, exporterhelper.WithRetry(cfg.BackOffConfig))
+	}
+
+	return options
+}
+
+func createTracesExporter(ctx context.Context, params exporter.Settings, cfg component.Config) (exporter.Traces, error) {
+	c := cfg.(*Config)
+	exporter, err := newTracesExporter(params, cfg)
+	if err != nil {
+		return nil, fmt.Errorf("cannot configure loadbalancing traces exporter: %w", err)
+	}
+
+	options := []exporterhelper.Option{
+		exporterhelper.WithStart(exporter.Start),
+		exporterhelper.WithShutdown(exporter.Shutdown),
+		exporterhelper.WithCapabilities(exporter.Capabilities()),
+	}
+
+	return exporterhelper.NewTraces(
+		ctx,
+		params,
+		cfg,
+		exporter.ConsumeTraces,
+		buildExporterResilienceOptions(options, c)...,
+	)
 }
 
-func createLogsExporter(_ context.Context, params exporter.Settings, cfg component.Config) (exporter.Logs, error) {
-	return newLogsExporter(params, cfg)
+func createLogsExporter(ctx context.Context, params exporter.Settings, cfg component.Config) (exporter.Logs, error) {
+	c := cfg.(*Config)
+	exporter, err := newLogsExporter(params, cfg)
+	if err != nil {
+		return nil, fmt.Errorf("cannot configure loadbalancing logs exporter: %w", err)
+	}
+
+	options := []exporterhelper.Option{
+		exporterhelper.WithStart(exporter.Start),
+		exporterhelper.WithShutdown(exporter.Shutdown),
+		exporterhelper.WithCapabilities(exporter.Capabilities()),
+	}
+
+	return exporterhelper.NewLogs(
+		ctx,
+		params,
+		cfg,
+		exporter.ConsumeLogs,
+		buildExporterResilienceOptions(options, c)...,
+	)
 }
 
-func createMetricsExporter(_ context.Context, params exporter.Settings, cfg component.Config) (exporter.Metrics, error) {
-	return newMetricsExporter(params, cfg)
+func createMetricsExporter(ctx context.Context, params exporter.Settings, cfg component.Config) (exporter.Metrics, error) {
+	c := cfg.(*Config)
+	exporter, err := newMetricsExporter(params, cfg)
+	if err != nil {
+		return nil, fmt.Errorf("cannot configure loadbalancing metrics exporter: %w", err)
+	}
+
+	options := []exporterhelper.Option{
+		exporterhelper.WithStart(exporter.Start),
+		exporterhelper.WithShutdown(exporter.Shutdown),
+		exporterhelper.WithCapabilities(exporter.Capabilities()),
+	}
+
+	return exporterhelper.NewMetrics(
+		ctx,
+		params,
+		cfg,
+		exporter.ConsumeMetrics,
+		buildExporterResilienceOptions(options, c)...,
+	)
 }