From: Ondřej Surý <ondrej@isc.org>
Date: Thu, 14 May 2026 16:18:27 +0000 (+0200)
Subject: Raise the per-server recursive-clients ceiling in fetchlimit
X-Git-Url: http://git.ipfire.org/gitweb/index.cgi?a=commitdiff_plain;h=db28b2127abdd92d83a1da8ae84b7b88dce53a86;p=thirdparty%2Fbind9.git

Raise the per-server recursive-clients ceiling in fetchlimit

With the resolver now legitimately escalating to TCP after repeated
UDP timeouts to the same authoritative, each lame-server lookup
takes ~50% longer to fail.  The recursive-client backlog therefore
peaks a little higher before the fetches-per-server auto-tune drops
the quota below 200.

Bump the upper bound for the burst-against-lame-server and recovery
steps from 200 to 250 to absorb that extra latency.  The lower bound
and the final post-recovery target (clients <= 20) are unchanged.

Assisted-by: Claude:claude-opus-4-7
---

diff --git a/bin/tests/system/fetchlimit/tests.sh b/bin/tests/system/fetchlimit/tests.sh
index 3d8a77397ab..b36f7925706 100644
--- a/bin/tests/system/fetchlimit/tests.sh
+++ b/bin/tests/system/fetchlimit/tests.sh
@@ -82,9 +82,11 @@ for try in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20; do
   burst 10.53.0.3 a $try
   # fetches-per-server is at 400, but at 20qps against a lame server,
   # we'll reach 200 at the tenth second, and the quota should have been
-  # tuned to less than that by then.
+  # tuned to less than that by then.  Allow a small margin above 200
+  # to absorb the extra latency introduced by the resolver's TCP
+  # fallback after repeated UDP timeouts.
   [ $try -le 5 ] && low=$((try * 10))
-  stat 10.53.0.3 20 200 || ret=1
+  stat 10.53.0.3 20 250 || ret=1
   [ $ret -eq 1 ] && break
   sleep 1
 done
@@ -125,7 +127,7 @@ ret=0
 sendcmd 10.53.0.4 send-responses "enable"
 for try in 1 2 3 4 5; do
   burst 10.53.0.3 b $try
-  stat 10.53.0.3 0 200 || ret=1
+  stat 10.53.0.3 0 250 || ret=1
   [ $ret -eq 1 ] && break
   sleep 1
 done