[Bug] INCLUDE (non-key) index column referenced by a predicate is treated as valid

Ouverte
  • Quelque chose ne va pas ?
#2,126 0 commentaires 0 réactions 0 personnes assignées Voir sur GitHub

Les mainteneurs répondent en général sous 3 jours

Personne n'a encore pris cette issue.

Évaluation

Difficulté
4/5
Temps estimé
3-5 jours
Accessibilité débutants
32/100
Type d'issue
Bug
Clarté
Plutôt claire
Activité
Active
Stack technique
c, postgresql
Domaine
databases

Piste de recherche

The bug is in GPORCA's construction of index conditions, and the issue suspects CXformUtils::PdrgpcrIndexColumns, which merges key and INCLUDE columns into one array. Reproduce the two SELECT statements with optimizer on and compare against optimizer off, then check which Index Cond the plan builds and where nodeIndexscan.c rejects it. Done means both queries return rows under optimizer=on without either sanity-check error. The issue reports an unmerged local fix but does not name a target design, so confirm the expected behaviour with maintainers first.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

type: Bug
Apache Cloudberry version

No response

What happened

A GPORCA bug: an INCLUDE (non-key) index column referenced by a predicate is treated as a valid
index key column when building an Index Cond (likely via CXformUtils::PdrgpcrIndexColumns, which
concatenates key and included columns into one array with no marker distinguishing the two). The
resulting plan carries a scan key against an attribute beyond indnkeyatts, which the executor
rejects. Which sanity check fires depends on the predicate, so one bug produces two error
messages -- both are suppressed below:

CREATE TABLE t(a int, b ); CREATE INDEX i ON t USING btree (a) INCLUDE (b);
SET optimizer = on;
SELECT a FROM t WHERE b; -- b boolean: "bogus index qualification (nodeIndexscan.c)"
SELECT a FROM t WHERE b IS NULL; -- b int: "btree index keys must be ordered by attribute"

Verified scope: optimizer=on only (the Postgres planner never builds such an Index Cond, and
every case above returns the correct rows under optimizer=off); not specific to a distribution
policy (DISTRIBUTED BY and DISTRIBUTED REPLICATED both reproduce); needs no data, an empty table
is enough; and the INCLUDE list does not have to overlap the key columns -- a plain
"INCLUDE (b)" is sufficient. A fix exists locally but has not been merged into the Cloudberry
main branch this environment is built from.

What you think should happen instead

No response

How to reproduce
CREATE TABLE t(a int, b int); CREATE INDEX i ON t USING btree (a) INCLUDE (b);
SET optimizer = on;
SELECT a FROM t WHERE b IS NULL; -- b int: "btree index keys must be ordered by attribute"
Operating System

any

Anything else

No response

Are you willing to submit PR?
  • Yes, I am willing to submit a PR!
Code of Conduct
Langage dominant
C
Étoiles
1.4k
Forks
265
Merge moyen
3 j 22 h
PR mergées (30 j)
82

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de apache/cloudberry

Toutes les issues de apache/cloudberry

Issues similaires

Plus d'issues C

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.