Fix assertion when decrementing eager scanning success and failure counters.
authorMasahiko Sawada <msawada@postgresql.org>
Tue, 27 May 2025 18:42:36 +0000 (11:42 -0700)
committerMasahiko Sawada <msawada@postgresql.org>
Tue, 27 May 2025 18:42:36 +0000 (11:42 -0700)
commit4c08ecd1618e3c5da664ba24a4aa7052772c4616
tree738c21c6d3264f294b347ab604ae6b22763e6a0f
parentc53f3b9cc8b1ece539291860a2def55ce1abf64a
Fix assertion when decrementing eager scanning success and failure counters.

Previously, we asserted that the eager scan's success and failure
counters were positive before decrementing them. However, this
assumption was incorrect, as it's possible that some blocks have
already been eagerly scanned by the time eager scanning is disabled.

This commit replaces the assertions with guards to handle this
scenario gracefully.

With this change, we continue to allow read-ahead operations by the
read stream that exceed the success and failure caps. While there is a
possibility that overruns will trigger eager scans of additional
pages, this does not pose a practical concern as the overruns will not
be substantial and remain within an acceptable range.

Reviewed-by: Melanie Plageman <melanieplageman@gmail.com>
Discussion: https://postgr.es/m/CAD21AoConf6tkVCv-=JhQJj56kYsDwo4jG5+WqgT+ukSkYomSQ@mail.gmail.com
src/backend/access/heap/vacuumlazy.c