From 0c7fa8d572bf3e742a627ff660175683e131908b Mon Sep 17 00:00:00 2001
From: Aram Sargsyan <aram@isc.org>
Date: Thu, 6 Mar 2025 14:37:09 +0000
Subject: [PATCH] Test resolver statistics when responses time out

Add a test to check that the timed out responses do not skew the
normal responses statistics counters, and that they do update the
timeouts counter.
---
 bin/tests/system/resolver/tests.sh            | 20 +++++++++++++++++++
 .../system/resolver/tests_sh_resolver.py      |  1 +
 2 files changed, 21 insertions(+)

diff --git a/bin/tests/system/resolver/tests.sh b/bin/tests/system/resolver/tests.sh
index 2669021c62..01de8e7d7f 100755
--- a/bin/tests/system/resolver/tests.sh
+++ b/bin/tests/system/resolver/tests.sh
@@ -43,6 +43,12 @@ grep "status: NOERROR" dig.out.ns1.test${n} >/dev/null || ret=1
 if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status + ret))
 
+rndccmd 10.53.0.1 stats || ret=1 # Get the responses, RTT and timeout statistics before the following timeout tests
+grep -F 'responses received' ns1/named.stats >ns1/named.stats.responses-before || true
+grep -F 'queries with RTT' ns1/named.stats >ns1/named.stats.rtt-before || true
+grep -F 'query timeouts' ns1/named.stats >ns1/named.stats.timeouts-before || true
+mv ns1/named.stats ns1/named.stats-before
+
 # 'resolver-query-timeout' is set to 5 seconds in ns1, so dig with a lower
 # timeout value should give up earlier than that.
 n=$((n + 1))
@@ -66,6 +72,20 @@ grep -F "EDE: 22 (No Reachable Authority)" dig.out.ns1.test${n} >/dev/null || re
 if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status + ret))
 
+n=$((n + 1))
+echo_i "checking that the timeout didn't skew the resolver responses counters and did update the timeout counter ($n)"
+ret=0
+rndccmd 10.53.0.1 stats || ret=1
+grep -F 'responses received' ns1/named.stats >ns1/named.stats.responses-after || true
+grep -F 'queries with RTT' ns1/named.stats >ns1/named.stats.rtt-after || true
+grep -F 'query timeouts' ns1/named.stats >ns1/named.stats.timeouts-after || true
+mv ns1/named.stats ns1/named.stats-after
+diff ns1/named.stats.responses-before ns1/named.stats.responses-after >/dev/null || ret=1
+diff ns1/named.stats.rtt-before ns1/named.stats.rtt-after >/dev/null || ret=1
+diff ns1/named.stats.timeouts-before ns1/named.stats.timeouts-after >/dev/null && ret=1
+if [ $ret != 0 ]; then echo_i "failed"; fi
+status=$((status + ret))
+
 # 'resolver-query-timeout' is set to 5 seconds in ns1, so named should
 # interrupt the non-responsive query and send a SERVFAIL answer before dig's
 # own timeout fires, which is set to 7 seconds. This time, exampleudp.net is
diff --git a/bin/tests/system/resolver/tests_sh_resolver.py b/bin/tests/system/resolver/tests_sh_resolver.py
index 6e5a7967df..10a36ce4e4 100644
--- a/bin/tests/system/resolver/tests_sh_resolver.py
+++ b/bin/tests/system/resolver/tests_sh_resolver.py
@@ -21,6 +21,7 @@ pytestmark = pytest.mark.extra_artifacts(
         "nextpart.out.*",
         "ans*/ans.run",
         "ans*/query.log",
+        "ns1/named.stats*",
         "ns4/tld.db",
         "ns5/trusted.conf",
         "ns6/K*",