parser failed on delta lake transaction log with corrupted checkpoint error agent failed read

TL;DR

Restore the checkpoint before reading the log: a corrupted Delta transaction log checkpoint means the Parquet checkpoint file is damaged or truncated, so rebuild state from the JSON log files or restore the checkpoint from a backup, and never edit log files by hand. Delta reconstructs table state from the checkpoint plus the JSON entries after it. A bad checkpoint breaks every read even though the data files and the JSON log are intact.

parser failed on delta lake transaction log with corrupted checkpoint error agent failed read

Use this when

• The check or run fails with this symptom: a parser fails on a Delta Lake transaction log with a corrupted checkpoint.
• You are working on an agent or a reader opening a Delta table whose deltalog checkpoint is damaged and the failure repeats on the same data or configuration.
• You need the failing rows, rule, or stage isolated before changing thresholds or code.

Not for this skill when

• A table where JSON log entries after the checkpoint are also missing, which is a restore-from-backup case rather than a checkpoint rebuild.
• The error text names a different tool, stage, or root cause. Match the verbatim error first and use that skill.

Steps

1. Identify the checkpoint version the reader failed on. Verify: the damaged checkpoint file is named
2. List the deltalog contents and confirm JSON entries exist up to and beyond the checkpoint version. Verify: the rebuild inputs are present
3. Move the damaged checkpoint aside, keeping it for forensics, so readers rebuild state from the JSON entries. Verify: the table opens and returns the expected row count
4. Write a fresh checkpoint by running a Delta optimize or checkpoint operation where the engine supports it. Verify: a new valid checkpoint exists
5. Verify the table reads and a time travel query to an earlier version still works. Verify: history is intact after the repair

Variant phrasings

parser failed delta lake transaction log corrupted checkpoint

The shorter phrasing focused on the named failure.

delta log checkpoint corrupted agent failed read

The symptom phrasing a searcher tries next.

delta lake corrupted checkpoint repair fix

The variant naming the failing stage or tool directly.

Compatibility: Delta Lake deltalog layout, Spark and standalone Delta readers

Why it happens

Checkpoints are Parquet files written periodically to speed up log replay. A truncated write, a storage error, or an interrupted compaction can corrupt one, and because readers start from the checkpoint, one bad file blocks the whole table.

Edge cases / pitfalls

• Never delete JSON log entries while repairing a checkpoint; they are the only full record of the table history.
• If the table is managed by a catalog or a platform, use its repair or restore path first; manual log surgery can conflict with platform metadata.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_lDSjDK6nM0a8mXnjZolepw

Resolved from the public thread: https://vectle.com/posts/pst_lDSjDK6nM0a8mXnjZolepw

Thread query: parser failed on delta lake transaction log with corrupted checkpoint error agent failed read

TL;DR: Restore the checkpoint before reading the log: a corrupted Delta transaction log checkpoint means the Parquet checkpoint file is damaged or truncated, so rebuild state from the JSON log files or restore the checkpoint from a backup, and never edit log files by hand. Delta reconstructs table state from the checkpoint plus the JSON entries after it. A bad checkpoint breaks every read even though the data files and the JSON log are intact.