Horton

Re: [PATCH] OAuth: fix performance bug with stuck multiplexer events

Jacob Champion <jacob.champion@enterprisedb.com>

View thread
From: Jacob Champion <jacob.champion@enterprisedb.com>
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Cc: Daniel Gustafsson <daniel@yesql.se>, Thomas Munro <thomas.munro@gmail.com>, Peter Eisentraut <peter@eisentraut.org>
Date: 2025-07-28T20:52:42Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. oauth: Always link with -lm for floor()

  2. oauth: Add unit tests for multiplexer handling

  3. oauth: Ensure unused socket registrations are removed

  4. oauth: Remove expired timers from the multiplexer

  5. oauth: Remove stale events from the kqueue multiplexer

  6. oauth: Track total call count during a client flow

Attachments

Hi all,

On Thu, Jun 26, 2025 at 4:33 PM Jacob Champion
<jacob.champion@enterprisedb.com> wrote:
> My plan, if this code seems reasonable, is to backport 0001-0003, but
> keep the larger 0004 on HEAD only until it has proven to be stable.
> It's a big new suite and I want to make sure it's not flapping on some
> buildfarm animal. Eventually I'll backport that too.

Any thoughts on the approach? Too big/too scary/too BSD-specific?

A small bit of self-review: a comment I wrote in the tests suggested
that the choice of readable/writable events was up to the multiplexer
implementation, but it *must* choose readable, due to the hardcoded
use of PGRES_POLLING_READING throughout the current code. Updated in
v2.

Thanks,
--Jacob