From: Olivier Houchard <ohouchard@haproxy.com>
Date: Thu, 30 Jul 2026 13:59:06 +0000 (+0200)
Subject: DOC: config: Document the tune.fd.tables option
X-Git-Url: http://git.ipfire.org/gitweb/?a=commitdiff_plain;h=0cec4be8cf833ccaccbfa947d43c053b42b71764;p=thirdparty%2Fhaproxy.git

DOC: config: Document the tune.fd.tables option

Add documentation for tune.fd.tables.
---

diff --git a/doc/configuration.txt b/doc/configuration.txt
index a8e5192b1..71e78b221 100644
--- a/doc/configuration.txt
+++ b/doc/configuration.txt
@@ -1891,6 +1891,7 @@ The following keywords are supported in the "global" section :
    - tune.events.max-events-at-once
    - tune.fail-alloc
    - tune.fd.edge-triggered
+   - tune.fd.tables
    - tune.h1.be.glitches-threshold
    - tune.h1.fe.glitches-threshold
    - tune.h1.zero-copy-fwd-recv
@@ -4306,6 +4307,14 @@ tune.fd.edge-triggered { on | off }  [ EXPERIMENTAL ]
   certain scenarios. This is still experimental, it may result in frozen
   connections if bugs are still present, and is disabled by default.
 
+tune.fd.tables { shared | per-thread-group }  [ EXPERIMENTAL ]
+  Decide if the file descriptor tables are shared by everybody (shared), or
+  if there is one per thread group (per-thread-group). Having one table per
+  thread group can give a good performance boost with enough threads and
+  thread groups, as there is a lot of contention on the kernel file descriptor
+  tables.
+  The default is "shared".
+
 tune.glitches.kill.cpu-usage <number>
   Sets the minimum CPU usage between 0 and 100, at which connections showing
   too many glitches will be killed. This applies to connections that have