hyper: a Parquet file with a footer over 16 MiB is refused

Aperta
  • Qualcosa non va?
#3,291 1 commento 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
4/5
Tempo stimato
3-5 giorni
Idoneità per principianti
15/100
Tipo di issue
Bug
Chiarezza
Specificata chiaramente
Stato di attività
Attiva
Stack tecnologico
shell
Ambito
databases

Direzione di ricerca

The failure comes from the Hyper engine's Parquet reader, not from the ClickBench scripts, so the scripts under hyper/load only show how the file is loaded. Start with the reproduction: write hits-small-groups.parquet with the ClickHouse command in the issue, then load it with hyper/load and confirm the 16777216-byte footer error. Done would mean the load succeeds, or the limit becomes configurable and hyper/load sets it. Because the limit is inside Hyper, the fix needs upstream engine changes and maintainer input.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

Hyper refuses a Parquet file whose footer is larger than 16 MiB:

tableauhyperapi.hyperexception.HyperException: error while reading Parquet file: Error in hits.parquet
	The footer size 35231477 of the Parquet file exceeds the maximum allowed footer size which is 16777216

This happened in the randomized benchmark (https://github.com/ClickHouse/ClickBench/pull/3271), which loads a generated dataset of a billion rows with the same schema as hits.parquet, using hyper/load unchanged. The file was written by ClickHouse 26.10 with its default row group size limit of 512 MB, which gives more than a thousand row groups for a billion rows, and their metadata makes a footer of 35 MB. The Parquet format has no such limit, and the other systems of the benchmark (ClickHouse, DuckDB, Firebolt, PostgreSQL-based ones, and others) loaded the same file.

Reproduction with the standard dataset: rewriting hits.parquet (100M rows, 226 row groups, a footer of 2.4 MB) with smaller row groups gives 2360 row groups and a footer of 31.6 MB:

clickhouse local --output_format_parquet_row_group_size 40000 --output_format_parquet_row_group_size_bytes 100000000000 \
    --query "SELECT * FROM file('hits.parquet') INTO OUTFILE 'hits-small-groups.parquet' TRUNCATE FORMAT Parquet"
clickhouse local --query "SELECT num_row_groups, metadata_size FROM file('hits-small-groups.parquet', ParquetMetadata)"
# 2360	31626081

Loading hits-small-groups.parquet with hyper/load should fail with the same message.

Could the limit be raised (or made configurable through an option that hyper/load could set)? The randomized benchmark now writes about 250 row groups per file, as the original file has, so its next runs avoid the limit, but files with many row groups are common for large datasets.

Hyper API version in the run: tableauhyperapi 0.0.26963 (hyper/install of main on 2026-11-13), machine c6a.metal.

@caetanosauer @vogelsgesang

🤖 Generated with Claude Code

Lingua principale
Shell
Stelle
1.2k
Fork
334
Merge medio
3h 50m
PR unite (30g)
710

Preparare l'ambiente

  • Nessun Dockerfile né file Docker Compose
  • Ha un modello di pull request
  • Nessuna guida per i contributori

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di ClickHouse/ClickBench

Tutte le issue di ClickHouse/ClickBench

Issue simili

Altre issue su Shell/Bash

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.