> On Mar 13, 2021, at 11:06 AM, Peter Geoghegan wrote:
>
> On Sat, Mar 13, 2021 at 10:35 AM Mark Dilger
> wrote:
>> The testing strategy I'm using is to corrupt heap and btree pages in schema
>> "public" of the "regression" database created by `make installcheck`, by
>> overwriting random b
On Sat, Mar 13, 2021 at 10:35 AM Mark Dilger
wrote:
> The testing strategy I'm using is to corrupt heap and btree pages in schema
> "public" of the "regression" database created by `make installcheck`, by
> overwriting random bytes in randomly selected locations on pages after the
> page header
Peter,
I'd like your advice on the following observations, if you'd be so kind:
Using the pg_amcheck command committed yesterday (thanks, Robert! thanks Tom!),
I have begun investigating segfaults that sometimes occur when running the
amcheck routines on intentionally corrupted databases. We'
