Re: Why is parula failing?

On 4/13/24 15:02, Robins Tharakan wrote:
> On Wed, 10 Apr 2024 at 10:24, David Rowley <dgrowleyml(at)gmail(dot)com> wrote:
>>
>> Master failed today for the first time since the compiler upgrade.
>> Again reltuples == 48.
>
> Here's what I can add over the past few days:
> - Almost all failures are either reltuples=48 or SIGABRTs
> - Almost all SIGABRTs are DDLs - CREATE INDEX / CREATE AGGREGATEs / CTAS
> - A little too coincidental? Recent crashes have stack-trace if
> interested.
>
> Barring the initial failures (during move to gcc 13.2), in the past week:
> - v15 somehow hasn't had a failure yet
> - v14 / v16 have got only 1 failure each
> - but v12 / v13 are lit up - failed multiple times.
>

I happened to have an unused rpi5, so I installed Armbian aarch64 with
gcc 13.2.0, built with exactly the same configure options as parula, and
did ~300 loops of "make check" without a single failure.

So either parula has packages in a different way, or maybe it's a more
of a timing issue and rpi5 is way slower than graviton3.

regards

--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company