ArchitectureA supervised JVM-class runtime — OLTP on seven engines, OLAP on three. AI-native, MCP-native, observable as plain SQL.Read the architecture
Está viendo la edición Perú. Está viendo la edición Colombia. You're viewing the Pakistan edition. Cambiar a la edición global →Cambiar a la edición global →Switch to the global edition →
Engineering

JVM classloader topology exposed as navigable observability tree

The JVM classloader topology is now queryable as a navigable catalog tree: a root table for the parent loader and three child detail tables covering loaded classes, classpath URLs, and per-loader resource counts — using the same browser pattern as the cache and schema catalogs.

Two new virtual-table groups expose the JVM classloader hierarchy inside the platform's observability catalog. The parent loader and each child loader appear as rows in the root table; drilling into a row returns the loader's loaded-class list, its classpath URL entries, and per-loader resource counts from dedicated child tables — the same navigation pattern already used for cache and schema catalogs, so operators work with a single mental model across all three surfaces.

What the tree exposes

  • Parent and child loader rows. The bootstrap and platform classloaders appear at the root alongside any child loaders registered by the platform's isolated-library mechanism, each with its type, delegation parent, and a live count of loaded classes.
  • Three child detail tables per loader. Loaded-class names with module and source-JAR coordinates; classpath URL entries in resolution order; and an aggregate row-count summary. Each table is reachable by drilling into the parent row from the catalog browser.
  • Catalog-style navigation. The structure mirrors the cache.catalog tree — the same browser that navigates cache entries navigates classloader state without additional tooling.

Diagnostic applications

  • Class-loading conflicts surface immediately. When two child loaders hold different versions of the same class, both rows appear in the loaded-class detail with their respective source JARs — no heap dump required.
  • Classpath drift detection. Operators diff the URL list between deployments to confirm which JARs changed without inspecting the filesystem directly.
  • Consistent permission model. Classloader state is governed by the same credential and role model that governs data access; no separate JMX connection or console access is required.

The classloader tree is read-only and refreshes on every query against the virtual tables. Combined with the cache observability tree from the same sprint, operators can trace a slow-path request from its SQL execution plan through its cache misses to the classloader that resolved the library under load — all through the platform's SQL interface.

See the feature →

← All posts