Include metadata_registry in the E2E regression tests

.changelog/889.feature.md

@@ -0,0 +1 @@
+Enable metadata_registry in the E2E regression tests

analyzer/metadata_registry/metadata_registry.go

@@ -32,31 +32,36 @@ const MetadataRegistryAnalyzerName = "metadata_registry"
 type processor struct {
 	target storage.TargetStorage
 	logger *log.Logger
+
+	gitCfg       registry.GitConfig
+	mockLogoUrls bool
 }
 
 var _ item.ItemProcessor[struct{}] = (*processor)(nil)
 
 func NewAnalyzer(
-	cfg config.ItemBasedAnalyzerConfig,
+	cfg config.MetadataRegistryConfig,
 	target storage.TargetStorage,
 	logger *log.Logger,
 ) (analyzer.Analyzer, error) {
 	logger.Info("Starting metadata_registry analyzer")
 	if cfg.Interval == 0 {
 		cfg.Interval = 2 * time.Minute
 	}
-	if cfg.Interval < time.Minute {
-		return nil, fmt.Errorf("invalid interval %s, metadata registry interval must be at least 1 minute", cfg.Interval)
-	}
 	logger = logger.With("analyzer", MetadataRegistryAnalyzerName)
 	p := &processor{
-		target: target,
-		logger: logger,
+		target:       target,
+		logger:       logger,
+		gitCfg:       registry.NewGitConfig(),
+		mockLogoUrls: cfg.MockLogoUrls,
+	}
+	if cfg.RepositoryBranch != "" {
+		p.gitCfg.Branch = cfg.RepositoryBranch
 	}
 
 	return item.NewAnalyzer[struct{}](
 		MetadataRegistryAnalyzerName,
-		cfg,
+		cfg.ItemBasedAnalyzerConfig,
 		p,
 		target,
 		logger,
@@ -68,7 +73,7 @@ func (p *processor) GetItems(ctx context.Context, limit uint64) ([]struct{}, err
 }
 
 func (p *processor) ProcessItem(ctx context.Context, batch *storage.QueryBatch, item struct{}) error {
-	gp, err := registry.NewGitProvider(registry.NewGitConfig())
+	gp, err := registry.NewGitProvider(p.gitCfg)
 	if err != nil {
 		return fmt.Errorf("failed to create Git registry provider: %s", err)
 	}
@@ -91,6 +96,9 @@ func (p *processor) ProcessItem(ctx context.Context, batch *storage.QueryBatch,
 				return ctx.Err()
 			}
 		}
+		if p.mockLogoUrls {
+			logoUrl = "http:://e2e-tests-mock-static-logo-url"
+		}
 
 		batch.Queue(
 			queries.ConsensusEntityMetaUpsert,

cmd/analyzer/analyzer.go

@@ -612,7 +612,7 @@ func NewService(cfg *config.AnalysisConfig) (*Service, error) { //nolint:gocyclo
 	}
 	if cfg.Analyzers.MetadataRegistry != nil {
 		analyzers, err = addAnalyzer(analyzers, err, "" /*syncTag*/, func() (A, error) {
-			return metadata_registry.NewAnalyzer(cfg.Analyzers.MetadataRegistry.ItemBasedAnalyzerConfig, dbClient, logger)
+			return metadata_registry.NewAnalyzer(*cfg.Analyzers.MetadataRegistry, dbClient, logger)
 		})
 	}
 	if cfg.Analyzers.ValidatorStakingHistory != nil {

config/config.go

@@ -523,13 +523,20 @@ type EvmAbiAnalyzerConfig struct {
 // MetadataRegistryConfig is the configuration for the metadata registry analyzer.
 type MetadataRegistryConfig struct {
 	ItemBasedAnalyzerConfig `koanf:",squash"`
+
+	// RepositoryBranch is the branch of the metadata registry repository to fetch.
+	// If unset, the default (production) branch is used. This is useful for E2E tests,
+	// where we use the 'nexus-e2e' branch, which remains stable.
+	RepositoryBranch string `koanf:"repository_branch"`
+
+	// MockLogoUrls is a flag to use mock URLs instead of keybase fetched URLs for logos.
+	// Useful to ensure static data for E2E testing since logo URLs can be updated without
+	// changes to the registry.
+	MockLogoUrls bool `koanf:"mock_logo_urls"`
 }
 
 // Validate validates the configuration.
 func (cfg *MetadataRegistryConfig) Validate() error {
-	if cfg.Interval < time.Minute {
-		return fmt.Errorf("metadata registry interval must be at least 1 minute")
-	}
 	return nil
 }
 

tests/e2e_regression/damask/e2e_config_2.yml

@@ -21,9 +21,9 @@ analysis:
     evm_abi_emerald:               { stop_if_queue_empty_for: 10s } # Give evm_contract_verifier time to fetch ABIs first. The 10s has been enough in practice, but might need to be tuned in the future, espcially if the caching proxy has an empty cache.
     evm_contract_verifier_emerald: { stop_if_queue_empty_for: 1s, sourcify_server_url: http://localhost:9191 }
     validator_staking_history:     { from: 8_048_956, stop_if_queue_empty_for: 1s, max_backoff_time: 6s }
+    metadata_registry:             { interval: 5s, stop_if_queue_empty_for: 1s, repository_branch: "nexus-e2e", mock_logo_urls: true }
     # Some non-block analyzers are not tested in e2e regressions.
     # They are largely not worth the trouble as they do not interact with rest of the system much.
-    # metadata_registry: {} # Awkward to inject mock registry responses.
     # node_stats: {}        # Awkward to inject mock node response using the current paradigm (= response caching).
     # aggregate_stats: {}   # Awkward to make stop after a single run.
   storage: