Support setting default scrape limit for operator components

Author: dehaansa

docs/sources/shared/reference/components/prom-operator-scrape.md

@@ -6,6 +6,7 @@ headless: true
 
 | Name                       | Type       | Description                                                                                                                  | Default | Required |
 |----------------------------|------------|------------------------------------------------------------------------------------------------------------------------------|---------|----------|
+| `default_sample_limit`     | `int`      | The default maximum samples per scrape. Used as the default if the target resource doesn't provide a sample limit.           |         | no       |
 | `default_scrape_interval`  | `duration` | The default interval between scraping targets. Used as the default if the target resource doesn't provide a scrape interval. | `1m`    | no       |
 | `default_scrape_timeout`   | `duration` | The default timeout for scrape requests. Used as the default if the target resource doesn't provide a scrape timeout.        | `10s`   | no       |
 | `scrape_native_histograms` | `bool`     | Whether to scrape native histograms from targets.                                                                            | `false` | no       |

internal/component/prometheus/operator/configgen/config_gen.go

@@ -190,6 +190,10 @@ func (cg *ConfigGenerator) generateDefaultScrapeConfig() *config.ScrapeConfig {
 		c.ScrapeTimeout = model.Duration(opt.DefaultScrapeTimeout)
 	}
 
+	if opt.DefaultSampleLimit != 0 {
+		c.SampleLimit = opt.DefaultSampleLimit
+	}
+
 	if opt.ScrapeNativeHistograms {
 		c.ScrapeProtocols = config.DefaultProtoFirstScrapeProtocols
 	}

internal/component/prometheus/operator/configgen/config_gen_podmonitor_test.go

@@ -92,6 +92,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 					},
 				},
 				ScrapeNativeHistograms:         falsePtr,
+				SampleLimit:                    18,
 				AlwaysScrapeClassicHistograms:  falsePtr,
 				ConvertClassicHistogramsToNHCB: falsePtr,
 				MetricNameValidationScheme:     model.LegacyValidation,
@@ -155,6 +156,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 						},
 					},
 				},
+				SampleLimit:                    18,
 				ScrapeNativeHistograms:         falsePtr,
 				AlwaysScrapeClassicHistograms:  falsePtr,
 				ConvertClassicHistogramsToNHCB: falsePtr,
@@ -219,6 +221,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 						},
 					},
 				},
+				SampleLimit:                    18,
 				ScrapeNativeHistograms:         falsePtr,
 				AlwaysScrapeClassicHistograms:  falsePtr,
 				ConvertClassicHistogramsToNHCB: falsePtr,
@@ -281,6 +284,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 						},
 					},
 				},
+				SampleLimit:                    18,
 				ScrapeNativeHistograms:         falsePtr,
 				AlwaysScrapeClassicHistograms:  falsePtr,
 				ConvertClassicHistogramsToNHCB: falsePtr,
@@ -345,6 +349,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 						},
 					},
 				},
+				SampleLimit:                    18,
 				ScrapeNativeHistograms:         falsePtr,
 				AlwaysScrapeClassicHistograms:  falsePtr,
 				ConvertClassicHistogramsToNHCB: falsePtr,
@@ -538,6 +543,7 @@ func TestGeneratePodMonitorConfig(t *testing.T) {
 					DefaultScrapeInterval:  time.Hour,
 					DefaultScrapeTimeout:   42 * time.Second,
 					ScrapeNativeHistograms: false,
+					DefaultSampleLimit:     18,
 				},
 			}
 			cfg, err := cg.GeneratePodMonitorConfig(tc.m, tc.ep, 1)

internal/component/prometheus/operator/configgen/config_gen_test.go

@@ -450,6 +450,7 @@ func TestGenerateDefaultScrapeConfig(t *testing.T) {
 			scrapeOptions: operator.ScrapeOptions{
 				DefaultScrapeInterval: 30 * time.Second,
 				DefaultScrapeTimeout:  5 * time.Second,
+				DefaultSampleLimit:    100,
 			},
 			expectedInterval:         30 * time.Second,
 			expectedTimeout:          5 * time.Second,
@@ -466,6 +467,7 @@ func TestGenerateDefaultScrapeConfig(t *testing.T) {
 			assert.Equal(t, model.Duration(tt.expectedInterval), got.ScrapeInterval)
 			assert.Equal(t, model.Duration(tt.expectedTimeout), got.ScrapeTimeout)
 			assert.Equal(t, tt.expectedFallbackProtocol, got.ScrapeFallbackProtocol)
+			assert.Equal(t, tt.scrapeOptions.DefaultSampleLimit, got.SampleLimit)
 		})
 	}
 }

internal/component/prometheus/operator/types.go

@@ -50,6 +50,9 @@ type ScrapeOptions struct {
 	// DefaultScrapeTimeout is the default timeout to scrape targets.
 	DefaultScrapeTimeout time.Duration `alloy:"default_scrape_timeout,attr,optional"`
 
+	// DefaultSampleLimit is the default sample limit per scrape.
+	DefaultSampleLimit uint `alloy:"default_sample_limit,attr,optional"`
+
 	// ScrapeNativeHistograms enables scraping of Prometheus native histograms.
 	ScrapeNativeHistograms bool `alloy:"scrape_native_histograms,attr,optional"`
 }
@@ -58,6 +61,7 @@ func (s *ScrapeOptions) GlobalConfig() promconfig.GlobalConfig {
 	cfg := promconfig.DefaultGlobalConfig
 	cfg.ScrapeInterval = model.Duration(s.DefaultScrapeInterval)
 	cfg.ScrapeTimeout = model.Duration(s.DefaultScrapeTimeout)
+	cfg.SampleLimit = s.DefaultSampleLimit
 	// TODO: add support for choosing validation scheme: https://github.com/grafana/alloy/issues/4122
 	cfg.MetricNameValidationScheme = model.LegacyValidation
 	cfg.MetricNameEscapingScheme = model.EscapeUnderscores