Re: Non-reproducible AIO failure - Mailing list pgsql-hackers

From Andres Freund
Subject Re: Non-reproducible AIO failure
Date
Msg-id dckrqf7qga5c2rcjxwaybfznrm736fefhc6tj2bf6hqyvq6svd@q6y7whb2u6r5
Whole thread Raw
In response to Re: Non-reproducible AIO failure  (Konstantin Knizhnik <knizhnik@garret.ru>)
List pgsql-hackers
Hi,

On 2025-09-03 21:50:42 +0300, Konstantin Knizhnik wrote:
> On 03/09/2025 8:37 PM, Dmitry Mityugov wrote:
> Size of PgAioHandle is144 bytes. I wonder how critical for us is to save 9
> bytes for it (3 bytes vs 3 integers)?

Not that it makes that huge a difference, but due to alignment considerations
the size increase would be 12 bytes, not 9. Maybe that could be addressed by
more reordering though.


> Why not to use normal enums instead of bitfields and uint8 with type casts?

There are a lot of PgAioHandles when max_connections is large. In addition,
some workloads (e.g. network attached storage), one might need to increase
io_max_concurrency to actually be able to saturate storage.  An ~8% increase
in size isn't nothing when the baseline isn't small.

Greetings,

Andres Freund



pgsql-hackers by date:

Previous
From: Peter Geoghegan
Date:
Subject: Re: index prefetching
Next
From: Masahiko Sawada
Date:
Subject: Re: POC: enable logical decoding when wal_level = 'replica' without a server restart