Date: Tue, 19 Aug 2025 13:44:39 -0500
From: Nathan Bossart <nathandbossart@gmail.com>
To: Andres Freund <andres@anarazel.de>
Cc: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>,
	Sami Imseih <samimseih@gmail.com>,
	Rahila Syed <rahilasyed90@gmail.com>,
	pgsql-hackers <pgsql-hackers@postgresql.org>
Subject: Re: Improve LWLock tranche name visibility across backends
Message-ID: <aKTGF9Fv1ioXa9P5@nathan>
References: <aKL1w6tczBqsqLiJ@ip-10-97-1-34.eu-west-3.compute.internal>
 <CAA5RZ0uuWWL5z80KEs95Evy9CrQqnR6vY91BmFor5YmEf06j=w@mail.gmail.com>
 <CAA5RZ0vPcUb68h16ueg=A1R4ps=qU74jvG8N2xaX97DAK-E7_A@mail.gmail.com>
 <aKOeMUjcv9Gozyyg@nathan>
 <CAA5RZ0viNsKqjSuJ8fo74SJrf0uMi=igOn5WrtNYmVqYVh5h-w@mail.gmail.com>
 <aKQxUS6rtMzCkEia@ip-10-97-1-34.eu-west-3.compute.internal>
 <aKS0auEl4UUlt7XO@nathan>
 <34drxi6kshtzu5aerftjflrm6wfxxw54emyvy4gqemdpxxuhvm@gymw6its72o7>
 <aKTDB5mvSowmC0Lf@nathan>
 <5to6tftuml6nkas4jaaljfzecasvslxq3mumeslh74wsol4mzw@rgxpxxlqqwtf>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <5to6tftuml6nkas4jaaljfzecasvslxq3mumeslh74wsol4mzw@rgxpxxlqqwtf>
Archived-At: <https://www.postgresql.org/message-id/aKTGF9Fv1ioXa9P5%40nathan>
Precedence: bulk

On Tue, Aug 19, 2025 at 02:37:19PM -0400, Andres Freund wrote:
> On 2025-08-19 13:31:35 -0500, Nathan Bossart wrote:
>> On Tue, Aug 19, 2025 at 02:06:50PM -0400, Andres Freund wrote:
>> > Possibly stupid question - is it really worth having a dynamic structure here?
>> > The number of tranches is strictly bound, it seems like it'd be simpler to
>> > have an array of tranch nmes in shared memory.
>> 
>> Tranches can be allocated post-startup with LWLockNewTrancheId() (e.g.,
>> autoprewarm).
> 
> Sure, but we don't need to support a large number of tranches. Just make it,
> idk, 128 entries long. Adding a dynamically allocated dsm to every server
> seems like a waste - ever shared mapping makes fork / exit slower...

The other issue is that there's presently no limit on the length of a
tranche name registered via LWLockRegisterTranche().  Life would become
much simpler if we're willing to put a limit on both that and the number of
tranches, but thus far we've been trying to avoid it.

-- 
nathan