> On 18 Jun 2025, at 11:33, Arseniy Mukhin <arseniy.mukhin....@gmail.com> wrote:
>
> Interesting, I used btree check as reference when started
> writing brin check, and in btree check there 53
> ERRCODE_INDEX_CORRUPTED ereports and only 1 ERRCODE_DATA_CORRUPTED
> ereport. So it was very hard to do, but I managed to pick the wrong
> one. I wonder if this btree check ereport should also be changed to
> ERRCODE_INDEX_CORRUPTED?
It's there in a case of heapallindexes failure. I concur that
ERRCODE_INDEX_CORRUPTED is more appropriate in that case in verify_nbtree.c.
But I recollect Peter explained this code before somewhere in pgsql-hackers.
And the reasoning was something like "if you lack a tuple in unquie constraints
- it's almost certainly subsequent constrain violation and data loss". But I'm
not sure.
And I could not find this discussion in archives.
Best regards, Andrey Borodin.
