| From: | Chapman Flack <chap(at)anastigmatix(dot)net> |
|---|---|
| To: | Omar Kilani <omar(dot)kilani(at)gmail(dot)com>, David Rowley <dgrowleyml(at)gmail(dot)com> |
| Cc: | PostgreSQL Developers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Strangeness with UNIQUE indexes and UTF-8 |
| Date: | 2021-06-06 16:36:26 |
| Message-ID: | 60BCF98A.2080101@anastigmatix.net |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 06/06/21 11:08, Omar Kilani wrote:
> I'm running pg_verify_checksums on the cluster, but the database is
> many TB so it'll be a bit.
Index corruption because of a locale change would not be the sort of thing
checksums would detect. Entries would be put into the index in the correct
order according to the old collation. The same entries can be still there,
intact, just fine according to the checksums, only the new collation would
have put them in a different order. Index search algorithms that are fast,
because they assume the entries to be correctly ordered, will skip regions
of the index where the desired key "couldn't possibly be", and if that's
where the old ordering put it, it won't be found.
Regards,
-Chap
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2021-06-06 16:38:26 | Re: Since '2001-09-09 01:46:40'::timestamp microseconds are lost when extracting epoch |
| Previous Message | Justin Pryzby | 2021-06-06 16:35:31 | pg14b1 stuck in lazy_scan_prune/heap_page_prune of pg_statistic |