Modern Docker Management, Designed for Everyone
Go composecontainer-managementcontainersdockerdocker-composedocker-managementgoself-hostedsveltekittypescriptweb-ui
fix: set explicit gorm LRU cache TTL to avoid constantly rising heap memory (#3102)
Kyle Mendell96bbe1463dff
diff --git a/backend/internal/database/database.go b/backend/internal/database/database.goindex c54f475e1..63cd3921a 100644--- a/backend/internal/database/database.go+++ b/backend/internal/database/database.go@@ -39,6 +39,19 @@ const (legacyVersionTable = "schema_migrations")+// Prepared-statement cache bounds. GORM's PrepareStmt cache is a global LRU keyed+// by SQL text. When PrepareStmtMaxSize/PrepareStmtTTL are left at zero, GORM falls+// back to math.MaxInt entries with a 24h TTL, i.e. effectively unbounded. Because+// this codebase emits highly variable SQL (dynamic filter/sort/pagination and+// GORM's IN (?,?,...) slice expansion, whose placeholder count changes the query+// text), the cache — and the modernc.org/sqlite compiled statements it retains on+// the Go heap — grows steadily under normal use. Bounding size and TTL keeps hot+// queries prepared while evicting the long tail (evicted statements are closed).+const (+ preparedStmtMaxSize = 256+ preparedStmtTTL = 15 * time.Minute+)+var customGormLogger logger.Interfacefunc SetGormLogger(l logger.Interface) {@@ -122,6 +135,8 @@ func connectDatabaseInternal(ctx context.Context, databaseURL string) (*DB, erroreturn time.Now().UTC()},PrepareStmt: true,+ PrepareStmtMaxSize: preparedStmtMaxSize,+ PrepareStmtTTL: preparedStmtTTL,IgnoreRelationshipsWhenMigrating: true,})if err == nil {