Hi Ken and Adrian,
Thanks for the useful suggestions!
We've investigated a bit more on the collations and we're almost certain it is the RC:
For Debian 11.11:
user=# select 'a' > 'A'; ?column? ---------- f (1 row) user=# select 'a' < 'A'; ?column? ---------- t (1 row)
For Alpine 11.19:
user=# select 'a' > 'A'; ?column? ---------- t (1 row) user=# select 'a' < 'A'; ?column? ---------- f (1 row)
We'll fix the image issue and re-index the affected tables.
Regards,
De
On 7/22/23 10:11, De Lan wrote:
> Hi pgsql community,
>
> Recently we found in a postgresql 11.19.0 alpine table there are two
> rows with duplicate primary keys.
>
>
> *Our questions:*
>
> Any ideas on what might cause this behavior other than the collation? if
> it is a well-known issue in the pgsql community or it really is the
> coalition that's the root cause, do we have mitigation for this kind of
> issue from happening in future?
Don't blindly update containers, test first.
>
>
>
> De,
>
> Thanks!
>
>
>
--
Adrian Klaver
adrian.klaver@aklaver.com