Apache Iceberg Internals and Snapshots

Explain the role of the catalog entry and its metadata-location pointer when loading an Iceberg table.

Classify Iceberg’s metadata JSON files, manifest lists, manifests, data files, and delete files by purpose.

Trace the immutable file graph from a table metadata file to the files visible in one snapshot.

Explain what an Iceberg snapshot represents and how the current snapshot relates to table history.

Interpret the snapshot ID, parent snapshot ID, sequence number, timestamp, operation, summary, and manifest-list fields.

Interpret ADDED, EXISTING, and DELETED manifest-entry statuses without confusing them with immediate S3 object deletion.

Explain how partition specs and transforms—including bucket transforms—describe data layout without relying on directory names.

Explain how immutable metadata files and an atomic catalog-pointer update support optimistic Iceberg commits.

Construct a timeline showing how a backup and a concurrent snapshot commit can observe different table states.

Define the exact consistency boundary for backing up one pinned snapshot and identify the transitive files it requires.

Distinguish a harmless concurrent commit from races involving snapshot expiration, orphan-file cleanup, or changing backup scope.

Map the repository’s Iceberg entity classes and serialized fields to the corresponding objects in the Iceberg file graph.

Trace the backup code path from its HMS Thrift read through each metadata or S3 object fetch.

Document the backup service’s intended contract, consistency assumptions, and concurrency-sensitive read points.

Design a deterministic test and a candidate fix for a snapshot created while a backup is in progress.