Add stale-cache-enable option and disable serve-stable by default

The current serve-stale implementation in BIND 9 stores all received
records in the cache for a max-stale-ttl interval (default 12 hours).

This allows DNS operators to turn the serve-stale answers in an event of
large authoritative DNS outage.  The caching of the stale answers needs
to be enabled before the outage happens or the feature would be
otherwise useless.

The negative consequence of the default setting is the inevitable
cache-bloat that happens for every and each DNS operator running named.

In this MR, a new configuration option `stale-cache-enable` is
introduced that allows the operators to selectively enable or disable
the serve-stale feature of BIND 9 based on their decision.

The newly introduced option has been disabled by default,
e.g. serve-stale is disabled in the default configuration and has to be
enabled if required.

bin/named/config.c

@@ -196,6 +196,7 @@ options {\n\
 #	sortlist <none>\n\
 	stale-answer-enable false;\n\
 	stale-answer-ttl 1; /* 1 second */\n\
+	stale-cache-enable false;\n\
 	synth-from-dnssec no;\n\
 #	topology <none>\n\
 	transfer-format many-answers;\n\

bin/named/named.conf.rst

@@ -401,6 +401,7 @@ OPTIONS
   	stacksize ( default | unlimited | sizeval );
   	stale-answer-enable boolean;
   	stale-answer-ttl duration;
+  	stale-cache-enable boolean;
   	startup-notify-rate integer;
   	statistics-file quoted_string;
   	synth-from-dnssec boolean;
@@ -785,6 +786,7 @@ VIEW
   	sortlist { address_match_element; ... };
   	stale-answer-enable boolean;
   	stale-answer-ttl duration;
+  	stale-cache-enable boolean;
   	synth-from-dnssec boolean;
   	transfer-format ( many-answers | one-answer );
   	transfer-source ( ipv4_address | * ) [ port ( integer | * ) ] [

bin/named/server.c

@@ -3895,7 +3895,7 @@ configure_view(dns_view_t *view, dns_viewlist_t *viewlist, cfg_obj_t *config,
 	uint32_t max_cache_size_percent = 0;
 	size_t max_adb_size;
 	uint32_t lame_ttl, fail_ttl;
-	uint32_t max_stale_ttl;
+	uint32_t max_stale_ttl = 0;
 	dns_tsig_keyring_t *ring = NULL;
 	dns_view_t *pview = NULL; /* Production view */
 	isc_mem_t *cmctx = NULL, *hmctx = NULL;
@@ -4358,9 +4358,14 @@ configure_view(dns_view_t *view, dns_viewlist_t *viewlist, cfg_obj_t *config,
 	view->synthfromdnssec = cfg_obj_asboolean(obj);
 
 	obj = NULL;
-	result = named_config_get(maps, "max-stale-ttl", &obj);
+	result = named_config_get(maps, "stale-cache-enable", &obj);
 	INSIST(result == ISC_R_SUCCESS);
-	max_stale_ttl = ISC_MAX(cfg_obj_asduration(obj), 1);
+	if (cfg_obj_asboolean(obj)) {
+		obj = NULL;
+		result = named_config_get(maps, "max-stale-ttl", &obj);
+		INSIST(result == ISC_R_SUCCESS);
+		max_stale_ttl = ISC_MAX(cfg_obj_asduration(obj), 1);
+	}
 
 	obj = NULL;
 	result = named_config_get(maps, "stale-answer-enable", &obj);

bin/tests/system/serve-stale/ns1/named1.conf.in

@@ -30,6 +30,7 @@ options {
 	max-stale-ttl 3600;
 	stale-answer-ttl 2;
 	stale-answer-enable yes;
+	stale-cache-enable yes;
 	servfail-ttl 0;
 };
 

bin/tests/system/serve-stale/ns1/named2.conf.in

@@ -30,6 +30,7 @@ options {
 	max-stale-ttl 20;
 	stale-answer-ttl 3;
 	stale-answer-enable yes;
+	stale-cache-enable yes;
 	servfail-ttl 0;
 };
 

bin/tests/system/serve-stale/ns3/named.conf.in

@@ -28,7 +28,7 @@ options {
 	listen-on-v6 { none; };
 	recursion yes;
 	dump-file "named_dump3.db";
-	// This configuration has no serve-stale options set.
+	stale-cache-enable yes;
 };
 
 zone "." {

bin/tests/system/serve-stale/ns4/named.conf.in

@@ -29,6 +29,7 @@ options {
 	recursion yes;
 	dump-file "named_dump4.db";
 	stale-answer-enable no;
+	stale-cache-enable yes;
 };
 
 zone "." {

bin/tests/system/serve-stale/setup.sh

@@ -16,3 +16,4 @@ $SHELL clean.sh
 copy_setports ns1/named1.conf.in ns1/named.conf
 copy_setports ns3/named.conf.in ns3/named.conf
 copy_setports ns4/named.conf.in ns4/named.conf
+copy_setports ns5/named.conf.in ns5/named.conf

bin/tests/system/serve-stale/tests.sh

@@ -861,7 +861,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 #
-# Now test server with serve-stale disabled.
+# Now test server with serve-stale answers disabled.
 #
 echo_i "test server with serve-stale disabled"
 
@@ -875,7 +875,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "prime cache longttl.example (serve-stale disabled) ($n)"
+echo_i "prime cache longttl.example (serve-stale answers disabled) ($n)"
 ret=0
 $DIG -p ${PORT} @10.53.0.4 longttl.example TXT > dig.out.test$n
 grep "status: NOERROR" dig.out.test$n > /dev/null || ret=1
@@ -884,7 +884,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "prime cache data.example (serve-stale disabled) ($n)"
+echo_i "prime cache data.example (serve-stale answers disabled) ($n)"
 ret=0
 $DIG -p ${PORT} @10.53.0.4 data.example TXT > dig.out.test$n
 grep "status: NOERROR" dig.out.test$n > /dev/null || ret=1
@@ -894,7 +894,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "prime cache othertype.example (serve-stale disabled) ($n)"
+echo_i "prime cache othertype.example (serve-stale answers disabled) ($n)"
 ret=0
 $DIG -p ${PORT} @10.53.0.4 othertype.example CAA > dig.out.test$n
 grep "status: NOERROR" dig.out.test$n > /dev/null || ret=1
@@ -904,7 +904,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "prime cache nodata.example (serve-stale disabled) ($n)"
+echo_i "prime cache nodata.example (serve-stale answers disabled) ($n)"
 ret=0
 $DIG -p ${PORT} @10.53.0.4 nodata.example TXT > dig.out.test$n
 grep "status: NOERROR" dig.out.test$n > /dev/null || ret=1
@@ -914,7 +914,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "prime cache nxdomain.example (serve-stale disabled) ($n)"
+echo_i "prime cache nxdomain.example (serve-stale answers disabled) ($n)"
 ret=0
 $DIG -p ${PORT} @10.53.0.4 nxdomain.example TXT > dig.out.test$n
 grep "status: NXDOMAIN" dig.out.test$n > /dev/null || ret=1
@@ -924,7 +924,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "verify prime cache statistics (serve-stale disabled) ($n)"
+echo_i "verify prime cache statistics (serve-stale answers disabled) ($n)"
 ret=0
 rm -f ns4/named.stats
 $RNDCCMD 10.53.0.4 stats > /dev/null 2>&1
@@ -972,7 +972,7 @@ waitfile dig.out.test$((n+3))
 waitfile dig.out.test$((n+4))
 
 n=$((n+1))
-echo_i "check fail of data.example (serve-stale disabled) ($n)"
+echo_i "check fail of data.example (serve-stale answers disabled) ($n)"
 ret=0
 grep "status: SERVFAIL" dig.out.test$n > /dev/null || ret=1
 grep "ANSWER: 0," dig.out.test$n > /dev/null || ret=1
@@ -980,7 +980,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "check fail of othertype.example (serve-stale disabled) ($n)"
+echo_i "check fail of othertype.example (serve-stale answers disabled) ($n)"
 ret=0
 grep "status: SERVFAIL" dig.out.test$n > /dev/null || ret=1
 grep "ANSWER: 0," dig.out.test$n > /dev/null || ret=1
@@ -988,7 +988,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "check fail of nodata.example (serve-stale disabled) ($n)"
+echo_i "check fail of nodata.example (serve-stale answers disabled) ($n)"
 ret=0
 grep "status: SERVFAIL" dig.out.test$n > /dev/null || ret=1
 grep "ANSWER: 0," dig.out.test$n > /dev/null || ret=1
@@ -996,7 +996,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "check fail of nxdomain.example (serve-stale disabled) ($n)"
+echo_i "check fail of nxdomain.example (serve-stale answers disabled) ($n)"
 ret=0
 grep "status: SERVFAIL" dig.out.test$n > /dev/null || ret=1
 grep "ANSWER: 0," dig.out.test$n > /dev/null || ret=1
@@ -1004,7 +1004,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 status=$((status+ret))
 
 n=$((n+1))
-echo_i "verify stale cache statistics (serve-stale disabled) ($n)"
+echo_i "verify stale cache statistics (serve-stale answers disabled) ($n)"
 ret=0
 rm -f ns4/named.stats
 $RNDCCMD 10.53.0.4 stats > /dev/null 2>&1
@@ -1024,7 +1024,7 @@ if [ $ret != 0 ]; then echo_i "failed"; fi
 
 # Dump the cache.
 n=$((n+1))
-echo_i "dump the cache (serve-stale disabled) ($n)"
+echo_i "dump the cache (serve-stale answers disabled) ($n)"
 ret=0
 $RNDCCMD 10.53.0.4 dumpdb -cache > rndc.out.test$n 2>&1 || ret=1
 done=0
@@ -1049,7 +1049,7 @@ LASTWEEK=`TZ=UTC perl -e 'my $now = time();
         printf("%04d%02d%02d%02d%02d%02d", $y+1900, $mo+1, $d, $h, $m, $s);'`
 
 n=$((n+1))
-echo_i "mock the cache date to $LASTWEEK (serve-stale disabled) ($n)"
+echo_i "mock the cache date to $LASTWEEK (serve-stale answers disabled) ($n)"
 ret=0
 sed -E "s/DATE [0-9]{14}/DATE $LASTWEEK/g" ns4/named_dump4.db > ns4/named_dumpdb4.db.out || ret=1
 cp ns4/named_dumpdb4.db.out ns4/named_dumpdb4.db
@@ -1060,7 +1060,7 @@ echo_i "start ns4"
 start_server --noclean --restart --port ${PORT} serve-stale ns4
 
 n=$((n+1))
-echo_i "verify ancient cache statistics (serve-stale disabled) ($n)"
+echo_i "verify ancient cache statistics (serve-stale answers disabled) ($n)"
 ret=0
 rm -f ns4/named.stats
 $RNDCCMD 10.53.0.4 stats #> /dev/null 2>&1

doc/arm/reference.rst

@@ -1815,9 +1815,9 @@ Boolean Options
    The default is ``yes``.
 
 ``stale-answer-enable``
-   If ``yes``, enable the returning of "stale" cached answers when the name servers
+   If ``yes``, enable the returning of "stale" cached answers when the name
-   for a zone are not answering. The default is not to return stale
+   servers for a zone are not answering and the ``stale-cache-enable`` option is
-   answers.
+   also enabled. The default is not to return stale answers.
 
    Stale answers can also be enabled or disabled at runtime via
    ``rndc serve-stale on`` or ``rndc serve-stale off``; these override
@@ -1831,6 +1831,9 @@ Boolean Options
    Information about stale answers is logged under the ``serve-stale``
    log category.
 
+``stale-cache-enable``
+   If ``yes``, enable the retaining of "stale" cached answers.  Default ``no``.
+
 ``nocookie-udp-size``
    This sets the maximum size of UDP responses that are sent to queries
    without a valid server COOKIE. A value below 128 is silently
@@ -3246,15 +3249,20 @@ Tuning
    (such as NS and glue AAAA/A records) in the resolution process.
 
 ``max-stale-ttl``
-   If stale answers are enabled, ``max-stale-ttl`` sets the maximum time
+   If retaining stale RRsets in cache is enabled, and returning of stale cached
+   answers is also enabled, ``max-stale-ttl`` sets the maximum time
    for which the server retains records past their normal expiry to
    return them as stale records, when the servers for those records are
    not reachable. The default is 12 hours. The minimum allowed is 1
    second; a value of 0 is updated silently to 1 second.
 
-   For stale answers to be returned, they must be enabled, either in the
+   For stale answers to be returned, the retaining of them in cache must be
-   configuration file using ``stale-answer-enable`` or via
+   enabled via the configuration option ``stale-cache-enable``, and returning
-   ``rndc serve-stale on``.
+   cached answers must be enabled, either in the configuration file using the
+   ``stale-answer-enable`` option or by calling ``rndc serve-stale on``.
+
+   When ``stale-cache-enable`` is set to ``no``, setting the ``max-stale-ttl``
+   has no effect, the value of ``max-cache-ttl`` will be ``0`` in such case.
 
 ``resolver-nonbackoff-tries``
    This specifies how many retries occur before exponential backoff kicks in. The

doc/man/named.conf.5in

@@ -464,6 +464,7 @@ options {
       stacksize ( default | unlimited | sizeval );
       stale\-answer\-enable boolean;
       stale\-answer\-ttl duration;
+      stale\-cache\-enable boolean;
       startup\-notify\-rate integer;
       statistics\-file quoted_string;
       synth\-from\-dnssec boolean;
@@ -876,6 +877,7 @@ view string [ class ] {
       sortlist { address_match_element; ... };
       stale\-answer\-enable boolean;
       stale\-answer\-ttl duration;
+      stale\-cache\-enable boolean;
       synth\-from\-dnssec boolean;
       transfer\-format ( many\-answers | one\-answer );
       transfer\-source ( ipv4_address | * ) [ port ( integer | * ) ] [

doc/misc/options

@@ -365,6 +365,7 @@ options {
         stacksize ( default | unlimited | <sizeval> );
         stale-answer-enable <boolean>;
         stale-answer-ttl <duration>;
+        stale-cache-enable <boolean>;
         startup-notify-rate <integer>;
         statistics-file <quoted_string>;
         statistics-interval <integer>; // ancient
@@ -747,6 +748,7 @@ view <string> [ <class> ] {
         sortlist { <address_match_element>; ... };
         stale-answer-enable <boolean>;
         stale-answer-ttl <duration>;
+        stale-cache-enable <boolean>;
         suppress-initial-notify <boolean>; // not yet implemented
         synth-from-dnssec <boolean>;
         topology { <address_match_element>; ... }; // ancient

doc/misc/options.active

@@ -329,6 +329,7 @@ options {
         stacksize ( default | unlimited | <sizeval> );
         stale-answer-enable <boolean>;
         stale-answer-ttl <duration>;
+        stale-cache-enable <boolean>;
         startup-notify-rate <integer>;
         statistics-file <quoted_string>;
         synth-from-dnssec <boolean>;
@@ -676,6 +677,7 @@ view <string> [ <class> ] {
         sortlist { <address_match_element>; ... };
         stale-answer-enable <boolean>;
         stale-answer-ttl <duration>;
+        stale-cache-enable <boolean>;
         synth-from-dnssec <boolean>;
         transfer-format ( many-answers | one-answer );
         transfer-source ( <ipv4_address> | * ) [ port ( <integer> | * ) ] [

doc/misc/options.grammar.rst

@@ -258,6 +258,7 @@
   	stacksize ( default | unlimited | <sizeval> );
   	stale-answer-enable <boolean>;
   	stale-answer-ttl <duration>;
+  	stale-cache-enable <boolean>;
   	startup-notify-rate <integer>;
   	statistics-file <quoted_string>;
   	synth-from-dnssec <boolean>;

lib/isccfg/namedconf.c

@@ -2045,6 +2045,7 @@ static cfg_clausedef_t view_clauses[] = {
 	{ "sortlist", &cfg_type_bracketed_aml, 0 },
 	{ "stale-answer-enable", &cfg_type_boolean, 0 },
 	{ "stale-answer-ttl", &cfg_type_duration, 0 },
+	{ "stale-cache-enable", &cfg_type_boolean, 0 },
 	{ "suppress-initial-notify", &cfg_type_boolean, CFG_CLAUSEFLAG_NYI },
 	{ "synth-from-dnssec", &cfg_type_boolean, 0 },
 	{ "topology", &cfg_type_bracketed_aml, CFG_CLAUSEFLAG_ANCIENT },