1unbound.conf(5) unbound 1.4.20 unbound.conf(5)
2
6 unbound.conf - Unbound configuration file.
7
9 unbound.conf
10
12 unbound.conf is used to configure unbound(8). The file format has
13 attributes and values. Some attributes have attributes inside them.
14 The notation is: attribute: value.
15 Comments start with # and last to the end of line. Empty lines are
17 ignored as is whitespace at the beginning of a line.
18 The utility unbound-checkconf(8) can be used to check unbound.conf
20 prior to usage.
21
23 An example config file is shown below. Copy this to
24 /etc/unbound/unbound.conf and start the server with:
25 $ unbound -c /etc/unbound/unbound.conf
27 Most settings are the defaults. Stop the server with:
29 $ kill `cat /etc/unbound/unbound.pid`
31 Below is a minimal config file. The source distribution contains an
33 extensive example.conf file with all the options.
34 # unbound.conf(5) config file for unbound(8).
36 server:
37 directory: "/etc/unbound"
38 username: unbound
39 # make sure unbound can access entropy from inside the chroot.
40 # e.g. on linux the use these commands (on BSD, devfs(8) is used):
41 # mount --bind -n /dev/random /etc/unbound/dev/random
42 # and mount --bind -n /dev/log /etc/unbound/dev/log
43 chroot: "/etc/unbound"
44 # logfile: "/etc/unbound/unbound.log" #uncomment to use logfile.
45 pidfile: "/etc/unbound/unbound.pid"
46 # verbosity: 1 # uncomment and increase to get more logging.
47 # listen on all interfaces, answer queries from the local subnet.
48 interface: 0.0.0.0
49 interface: ::0
50 access-control: 10.0.0.0/8 allow
51 access-control: 2001:DB8::/64 allow
52
54 There must be whitespace between keywords. Attribute keywords end with
55 a colon ':'. An attribute is followed by its containing attributes, or
56 a value.
57 Files can be included using the include: directive. It can appear any‐
59 where, it accepts a single file name as argument. Processing continues
60 as if the text from the included file was copied into the config file
61 at that point. If also using chroot, using full path names for the
62 included files works, relative pathnames for the included names work if
63 the directory where the daemon is started equals its chroot/working
64 directory. Wildcards can be used to include multiple files, see
65 glob(7).
66 Server Options
68 These options are part of the server: clause.
69 verbosity: <number>
71 The verbosity number, level 0 means no verbosity, only errors.
72 Level 1 gives operational information. Level 2 gives detailed
73 operational information. Level 3 gives query level information,
74 output per query. Level 4 gives algorithm level information.
75 Level 5 logs client identification for cache misses. Default is
76 level 1. The verbosity can also be increased from the command‐
77 line, see unbound(8).
78 statistics-interval: <seconds>
80 The number of seconds between printing statistics to the log for
81 every thread. Disable with value 0 or "". Default is disabled.
82 The histogram statistics are only printed if replies were sent
83 during the statistics interval, requestlist statistics are
84 printed for every interval (but can be 0). This is because the
85 median calculation requires data to be present.
86 statistics-cumulative: <yes or no>
88 If enabled, statistics are cumulative since starting unbound,
89 without clearing the statistics counters after logging the sta‐
90 tistics. Default is no.
91 extended-statistics: <yes or no>
93 If enabled, extended statistics are printed from unbound-con‐
94 trol(8). Default is off, because keeping track of more statis‐
95 tics takes time. The counters are listed in unbound-control(8).
96 num-threads: <number>
98 The number of threads to create to serve clients. Use 1 for no
99 threading.
100 port: <port number>
102 The port number, default 53, on which the server responds to
103 queries.
104 interface: <ip address[@port]>
106 Interface to use to connect to the network. This interface is
107 listened to for queries from clients, and answers to clients are
108 given from it. Can be given multiple times to work on several
109 interfaces. If none are given the default is to listen to local‐
110 host. The interfaces are not changed on a reload (kill -HUP)
111 but only on restart. A port number can be specified with @port
112 (without spaces between interface and port number), if not spec‐
113 ified the default port (from port) is used.
114 interface-automatic: <yes or no>
116 Detect source interface on UDP queries and copy them to replies.
117 This feature is experimental, and needs support in your OS for
118 particular socket options. Default value is no.
119 outgoing-interface: <ip address>
121 Interface to use to connect to the network. This interface is
122 used to send queries to authoritative servers and receive their
123 replies. Can be given multiple times to work on several inter‐
124 faces. If none are given the default (all) is used. You can
125 specify the same interfaces in interface: and outgoing-inter‐
126 face: lines, the interfaces are then used for both purposes.
127 Outgoing queries are sent via a random outgoing interface to
128 counter spoofing.
129 outgoing-range: <number>
131 Number of ports to open. This number of file descriptors can be
132 opened per thread. Must be at least 1. Default depends on com‐
133 pile options. Larger numbers need extra resources from the oper‐
134 ating system. For performance a a very large value is best, use
135 libevent to make this possible.
136 outgoing-port-permit: <port number or range>
138 Permit unbound to open this port or range of ports for use to
139 send queries. A larger number of permitted outgoing ports
140 increases resilience against spoofing attempts. Make sure these
141 ports are not needed by other daemons. By default only ports
142 above 1024 that have not been assigned by IANA are used. Give a
143 port number or a range of the form "low-high", without spaces.
144 The outgoing-port-permit and outgoing-port-avoid statements are
146 processed in the line order of the config file, adding the per‐
147 mitted ports and subtracting the avoided ports from the set of
148 allowed ports. The processing starts with the non IANA allo‐
149 cated ports above 1024 in the set of allowed ports.
150 outgoing-port-avoid: <port number or range>
152 Do not permit unbound to open this port or range of ports for
153 use to send queries. Use this to make sure unbound does not grab
154 a port that another daemon needs. The port is avoided on all
155 outgoing interfaces, both IP4 and IP6. By default only ports
156 above 1024 that have not been assigned by IANA are used. Give a
157 port number or a range of the form "low-high", without spaces.
158 outgoing-num-tcp: <number>
160 Number of outgoing TCP buffers to allocate per thread. Default
161 is 10. If set to 0, or if do_tcp is "no", no TCP queries to
162 authoritative servers are done.
163 incoming-num-tcp: <number>
165 Number of incoming TCP buffers to allocate per thread. Default
166 is 10. If set to 0, or if do_tcp is "no", no TCP queries from
167 clients are accepted.
168 edns-buffer-size: <number>
170 Number of bytes size to advertise as the EDNS reassembly buffer
171 size. This is the value put into datagrams over UDP towards
172 peers. The actual buffer size is determined by msg-buffer-size
173 (both for TCP and UDP). Do not set higher than that value.
174 Default is 4096 which is RFC recommended. If you have fragmen‐
175 tation reassembly problems, usually seen as timeouts, then a
176 value of 1480 can fix it. Setting to 512 bypasses even the most
177 stringent path MTU problems, but is seen as extreme, since the
178 amount of TCP fallback generated is excessive (probably also for
179 this resolver, consider tuning the outgoing tcp number).
180 msg-buffer-size: <number>
182 Number of bytes size of the message buffers. Default is 65552
183 bytes, enough for 64 Kb packets, the maximum DNS message size.
184 No message larger than this can be sent or received. Can be
185 reduced to use less memory, but some requests for DNS data, such
186 as for huge resource records, will result in a SERVFAIL reply to
187 the client.
188 msg-cache-size: <number>
190 Number of bytes size of the message cache. Default is 4
191 megabytes. A plain number is in bytes, append 'k', 'm' or 'g'
192 for kilobytes, megabytes or gigabytes (1024*1024 bytes in a
193 megabyte).
194 msg-cache-slabs: <number>
196 Number of slabs in the message cache. Slabs reduce lock con‐
197 tention by threads. Must be set to a power of 2. Setting
198 (close) to the number of cpus is a reasonable guess.
199 num-queries-per-thread: <number>
201 The number of queries that every thread will service simultane‐
202 ously. If more queries arrive that need servicing, and no
203 queries can be jostled out (see jostle-timeout), then the
204 queries are dropped. This forces the client to resend after a
205 timeout; allowing the server time to work on the existing
206 queries. Default depends on compile options, 512 or 1024.
207 jostle-timeout: <msec>
209 Timeout used when the server is very busy. Set to a value that
210 usually results in one roundtrip to the authority servers. If
211 too many queries arrive, then 50% of the queries are allowed to
212 run to completion, and the other 50% are replaced with the new
213 incoming query if they have already spent more than their
214 allowed time. This protects against denial of service by slow
215 queries or high query rates. Default 200 milliseconds. The
216 effect is that the qps for long-lasting queries is about (num‐
217 queriesperthread / 2) / (average time for such long queries)
218 qps. The qps for short queries can be about (numqueries‐
219 perthread / 2) / (jostletimeout in whole seconds) qps per
220 thread, about (1024/2)*5 = 2560 qps by default.
221 so-rcvbuf: <number>
223 If not 0, then set the SO_RCVBUF socket option to get more buf‐
224 fer space on UDP port 53 incoming queries. So that short spikes
225 on busy servers do not drop packets (see counter in netstat
226 -su). Default is 0 (use system value). Otherwise, the number
227 of bytes to ask for, try "4m" on a busy server. The OS caps it
228 at a maximum, on linux unbound needs root permission to bypass
229 the limit, or the admin can use sysctl net.core.rmem_max. On
230 BSD change kern.ipc.maxsockbuf in /etc/sysctl.conf. On OpenBSD
231 change header and recompile kernel. On Solaris ndd -set /dev/udp
232 udp_max_buf 8388608.
233 so-sndbuf: <number>
235 If not 0, then set the SO_SNDBUF socket option to get more buf‐
236 fer space on UDP port 53 outgoing queries. This for very busy
237 servers handles spikes in answer traffic, otherwise 'send:
238 resource temporarily unavailable' can get logged, the buffer
239 overrun is also visible by netstat -su. Default is 0 (use sys‐
240 tem value). Specify the number of bytes to ask for, try "4m" on
241 a very busy server. The OS caps it at a maximum, on linux
242 unbound needs root permission to bypass the limit, or the admin
243 can use sysctl net.core.wmem_max. On BSD, Solaris changes are
244 similar to so-rcvbuf.
245 rrset-cache-size: <number>
247 Number of bytes size of the RRset cache. Default is 4 megabytes.
248 A plain number is in bytes, append 'k', 'm' or 'g' for kilo‐
249 bytes, megabytes or gigabytes (1024*1024 bytes in a megabyte).
250 rrset-cache-slabs: <number>
252 Number of slabs in the RRset cache. Slabs reduce lock contention
253 by threads. Must be set to a power of 2.
254 cache-max-ttl: <seconds>
256 Time to live maximum for RRsets and messages in the cache.
257 Default is 86400 seconds (1 day). If the maximum kicks in,
258 responses to clients still get decrementing TTLs based on the
259 original (larger) values. When the internal TTL expires, the
260 cache item has expired. Can be set lower to force the resolver
261 to query for data often, and not trust (very large) TTL values.
262 cache-min-ttl: <seconds>
264 Time to live minimum for RRsets and messages in the cache.
265 Default is 0. If the the minimum kicks in, the data is cached
266 for longer than the domain owner intended, and thus less queries
267 are made to look up the data. Zero makes sure the data in the
268 cache is as the domain owner intended, higher values, especially
269 more than an hour or so, can lead to trouble as the data in the
270 cache does not match up with the actual data any more.
271 infra-host-ttl: <seconds>
273 Time to live for entries in the host cache. The host cache con‐
274 tains roundtrip timing, lameness and EDNS support information.
275 Default is 900.
276 infra-cache-slabs: <number>
278 Number of slabs in the infrastructure cache. Slabs reduce lock
279 contention by threads. Must be set to a power of 2.
280 infra-cache-numhosts: <number>
282 Number of hosts for which information is cached. Default is
283 10000.
284 do-ip4: <yes or no>
286 Enable or disable whether ip4 queries are answered or issued.
287 Default is yes.
288 do-ip6: <yes or no>
290 Enable or disable whether ip6 queries are answered or issued.
291 Default is yes. If disabled, queries are not answered on IPv6,
292 and queries are not sent on IPv6 to the internet nameservers.
293 do-udp: <yes or no>
295 Enable or disable whether UDP queries are answered or issued.
296 Default is yes.
297 do-tcp: <yes or no>
299 Enable or disable whether TCP queries are answered or issued.
300 Default is yes.
301 tcp-upstream: <yes or no>
303 Enable or disable whether the upstream queries use TCP only for
304 transport. Default is no. Useful in tunneling scenarios.
305 ssl-upstream: <yes or no>
307 Enabled or disable whether the upstream queries use SSL only for
308 transport. Default is no. Useful in tunneling scenarios. The
309 SSL contains plain DNS in TCP wireformat. The other server must
310 support this (see ssl-service-key).
311 ssl-service-key: <file>
313 If enabled, the server provider SSL service on its TCP sockets.
314 The clients have to use ssl-upstream: yes. The file is the pri‐
315 vate key for the TLS session. The public certificate is in the
316 ssl-service-pem file. Default is "", turned off. Requires a
317 restart (a reload is not enough) if changed, because the private
318 key is read while root permissions are held and before chroot
319 (if any). Normal DNS TCP service is not provided and gives
320 errors, this service is best run with a different port: config
321 or @port suffixes in the interface config.
322 ssl-service-pem: <file>
324 The public key certificate pem file for the ssl service.
325 Default is "", turned off.
326 ssl-port: <number>
328 The port number on which to provide TCP SSL service, default
329 443, only interfaces configured with that port number as @number
330 get the SSL service.
331 do-daemonize: <yes or no>
333 Enable or disable whether the unbound server forks into the
334 background as a daemon. Default is yes.
335 access-control: <IP netblock> <action>
337 The netblock is given as an IP4 or IP6 address with /size
338 appended for a classless network block. The action can be deny,
339 refuse, allow or allow_snoop.
340 The action deny stops queries from hosts from that netblock.
342 The action refuse stops queries too, but sends a DNS rcode
344 REFUSED error message back.
345 The action allow gives access to clients from that netblock. It
347 gives only access for recursion clients (which is what almost
348 all clients need). Nonrecursive queries are refused.
349 The allow action does allow nonrecursive queries to access the
351 local-data that is configured. The reason is that this does not
352 involve the unbound server recursive lookup algorithm, and
353 static data is served in the reply. This supports normal opera‐
354 tions where nonrecursive queries are made for the authoritative
355 data. For nonrecursive queries any replies from the dynamic
356 cache are refused.
357 The action allow_snoop gives nonrecursive access too. This give
359 both recursive and non recursive access. The name allow_snoop
360 refers to cache snooping, a technique to use nonrecursive
361 queries to examine the cache contents (for malicious acts).
362 However, nonrecursive queries can also be a valuable debugging
363 tool (when you want to examine the cache contents). In that case
364 use allow_snoop for your administration host.
365 By default only localhost is allowed, the rest is refused. The
367 default is refused, because that is protocol-friendly. The DNS
368 protocol is not designed to handle dropped packets due to pol‐
369 icy, and dropping may result in (possibly excessive) retried
370 queries.
371 chroot: <directory>
373 If chroot is enabled, you should pass the configfile (from the
374 commandline) as a full path from the original root. After the
375 chroot has been performed the now defunct portion of the config
376 file path is removed to be able to reread the config after a
377 reload.
378 All other file paths (working dir, logfile, roothints, and key
380 files) can be specified in several ways: as an absolute path
381 relative to the new root, as a relative path to the working
382 directory, or as an absolute path relative to the original root.
383 In the last case the path is adjusted to remove the unused por‐
384 tion.
385 The pidfile can be either a relative path to the working direc‐
387 tory, or an absolute path relative to the original root. It is
388 written just prior to chroot and dropping permissions. This
389 allows the pidfile to be /var/run/unbound.pid and the chroot to
390 be /var/unbound, for example.
391 Additionally, unbound may need to access /dev/random (for
393 entropy) from inside the chroot.
394 If given a chroot is done to the given directory. The default is
396 "/etc/unbound". If you give "" no chroot is performed.
397 username: <name>
399 If given, after binding the port the user privileges are
400 dropped. Default is "unbound". If you give username: "" no user
401 change is performed.
402 If this user is not capable of binding the port, reloads (by
404 signal HUP) will still retain the opened ports. If you change
405 the port number in the config file, and that new port number
406 requires privileges, then a reload will fail; a restart is
407 needed.
408 directory: <directory>
410 Sets the working directory for the program. Default is
411 "/etc/unbound".
412 logfile: <filename>
414 If "" is given, logging goes to stderr, or nowhere once daemo‐
415 nized. The logfile is appended to, in the following format:
416 [seconds since 1970] unbound[pid:tid]: type: message.
417 If this option is given, the use-syslog is option is set to
418 "no". The logfile is reopened (for append) when the config file
419 is reread, on SIGHUP.
420 use-syslog: <yes or no>
422 Sets unbound to send log messages to the syslogd, using sys‐
423 log(3). The log facility LOG_DAEMON is used, with identity
424 "unbound". The logfile setting is overridden when use-syslog is
425 turned on. The default is to log to syslog.
426 log-time-ascii: <yes or no>
428 Sets logfile lines to use a timestamp in UTC ascii. Default is
429 no, which prints the seconds since 1970 in brackets. No effect
430 if using syslog, in that case syslog formats the timestamp
431 printed into the log files.
432 log-queries: <yes or no>
434 Prints one line per query to the log, with the log timestamp and
435 IP address, name, type and class. Default is no. Note that it
436 takes time to print these lines which makes the server (signifi‐
437 cantly) slower. Odd (nonprintable) characters in names are
438 printed as '?'.
439 pidfile: <filename>
441 The process id is written to the file. Default is
442 "/var/run/unbound/unbound.pid". So,
443 kill -HUP `cat /var/run/unbound/unbound.pid`
444 triggers a reload,
445 kill -QUIT `cat /var/run/unbound/unbound.pid`
446 gracefully terminates.
447 root-hints: <filename>
449 Read the root hints from this file. Default is nothing, using
450 builtin hints for the IN class. The file has the format of zone
451 files, with root nameserver names and addresses only. The
452 default may become outdated, when servers change, therefore it
453 is good practice to use a root-hints file.
454 hide-identity: <yes or no>
456 If enabled id.server and hostname.bind queries are refused.
457 identity: <string>
459 Set the identity to report. If set to "", the default, then the
460 hostname of the server is returned.
461 hide-version: <yes or no>
463 If enabled version.server and version.bind queries are refused.
464 version: <string>
466 Set the version to report. If set to "", the default, then the
467 package version is returned.
468 target-fetch-policy: <"list of numbers">
470 Set the target fetch policy used by unbound to determine if it
471 should fetch nameserver target addresses opportunistically. The
472 policy is described per dependency depth.
473 The number of values determines the maximum dependency depth
475 that unbound will pursue in answering a query. A value of -1
476 means to fetch all targets opportunistically for that dependency
477 depth. A value of 0 means to fetch on demand only. A positive
478 value fetches that many targets opportunistically.
479 Enclose the list between quotes ("") and put spaces between num‐
481 bers. The default is "3 2 1 0 0". Setting all zeroes, "0 0 0 0
482 0" gives behaviour closer to that of BIND 9, while setting "-1
483 -1 -1 -1 -1" gives behaviour rumoured to be closer to that of
484 BIND 8.
485 harden-short-bufsize: <yes or no>
487 Very small EDNS buffer sizes from queries are ignored. Default
488 is off, since it is legal protocol wise to send these, and
489 unbound tries to give very small answers to these queries, where
490 possible.
491 harden-large-queries: <yes or no>
493 Very large queries are ignored. Default is off, since it is
494 legal protocol wise to send these, and could be necessary for
495 operation if TSIG or EDNS payload is very large.
496 harden-glue: <yes or no>
498 Will trust glue only if it is within the servers authority.
499 Default is on.
500 harden-dnssec-stripped: <yes or no>
502 Require DNSSEC data for trust-anchored zones, if such data is
503 absent, the zone becomes bogus. If turned off, and no DNSSEC
504 data is received (or the DNSKEY data fails to validate), then
505 the zone is made insecure, this behaves like there is no trust
506 anchor. You could turn this off if you are sometimes behind an
507 intrusive firewall (of some sort) that removes DNSSEC data from
508 packets, or a zone changes from signed to unsigned to badly
509 signed often. If turned off you run the risk of a downgrade
510 attack that disables security for a zone. Default is on.
511 harden-below-nxdomain: <yes or no>
513 From draft-vixie-dnsext-resimprove, returns nxdomain to queries
514 for a name below another name that is already known to be nxdo‐
515 main. DNSSEC mandates noerror for empty nonterminals, hence
516 this is possible. Very old software might return nxdomain for
517 empty nonterminals (that usually happen for reverse IP address
518 lookups), and thus may be incompatible with this. To try to
519 avoid this only DNSSEC-secure nxdomains are used, because the
520 old software does not have DNSSEC. Default is off.
521 harden-referral-path: <yes or no>
523 Harden the referral path by performing additional queries for
524 infrastructure data. Validates the replies if trust anchors are
525 configured and the zones are signed. This enforces DNSSEC vali‐
526 dation on nameserver NS sets and the nameserver addresses that
527 are encountered on the referral path to the answer. Default
528 off, because it burdens the authority servers, and it is not RFC
529 standard, and could lead to performance problems because of the
530 extra query load that is generated. Experimental option. If
531 you enable it consider adding more numbers after the tar‐
532 get-fetch-policy to increase the max depth that is checked to.
533 use-caps-for-id: <yes or no>
535 Use 0x20-encoded random bits in the query to foil spoof
536 attempts. This perturbs the lowercase and uppercase of query
537 names sent to authority servers and checks if the reply still
538 has the correct casing. Disabled by default. This feature is
539 an experimental implementation of draft dns-0x20.
540 private-address: <IP address or subnet>
542 Give IPv4 of IPv6 addresses or classless subnets. These are
543 addresses on your private network, and are not allowed to be
544 returned for public internet names. Any occurence of such
545 addresses are removed from DNS answers. Additionally, the DNSSEC
546 validator may mark the answers bogus. This protects against
547 so-called DNS Rebinding, where a user browser is turned into a
548 network proxy, allowing remote access through the browser to
549 other parts of your private network. Some names can be allowed
550 to contain your private addresses, by default all the local-data
551 that you configured is allowed to, and you can specify addi‐
552 tional names using private-domain. No private addresses are
553 enabled by default. We consider to enable this for the RFC1918
554 private IP address space by default in later releases. That
555 would enable private addresses for 10.0.0.0/8 172.16.0.0/12
556 192.168.0.0/16 169.254.0.0/16 fd00::/8 and fe80::/10, since the
557 RFC standards say these addresses should not be visible on the
558 public internet. Turning on 127.0.0.0/8 would hinder many spam‐
559 blocklists as they use that.
560 private-domain: <domain name>
562 Allow this domain, and all its subdomains to contain private
563 addresses. Give multiple times to allow multiple domain names
564 to contain private addresses. Default is none.
565 unwanted-reply-threshold: <number>
567 If set, a total number of unwanted replies is kept track of in
568 every thread. When it reaches the threshold, a defensive action
569 is taken and a warning is printed to the log. The defensive
570 action is to clear the rrset and message caches, hopefully
571 flushing away any poison. A value of 10 million is suggested.
572 Default is 0 (turned off).
573 do-not-query-address: <IP address>
575 Do not query the given IP address. Can be IP4 or IP6. Append
576 /num to indicate a classless delegation netblock, for example
577 like 10.2.3.4/24 or 2001::11/64.
578 do-not-query-localhost: <yes or no>
580 If yes, localhost is added to the do-not-query-address entries,
581 both IP6 ::1 and IP4 127.0.0.1/8. If no, then localhost can be
582 used to send queries to. Default is yes.
583 prefetch: <yes or no>
585 If yes, message cache elements are prefetched before they expire
586 to keep the cache up to date. Default is no. Turning it on
587 gives about 10 percent more traffic and load on the machine, but
588 popular items do not expire from the cache.
589 prefetch-key: <yes or no>
591 If yes, fetch the DNSKEYs earlier in the validation process,
592 when a DS record is encountered. This lowers the latency of
593 requests. It does use a little more CPU. Also if the cache is
594 set to 0, it is no use. Default is no.
595 rrset-roundrobin: <yes or no>
597 If yes, Unbound rotates RRSet order in response (the random num‐
598 ber is taken from the query ID, for speed and thread safety).
599 Default is no.
600 minimal-responses: <yes or no>
602 If yes, Unbound doesn't insert authority/additional sections
603 into response messages when those sections are not required.
604 This reduces response size significantly, and may avoid TCP
605 fallback for some responses. This may cause a slight speedup.
606 The default is no, because the DNS protocol RFCs mandate these
607 sections, and the additional content could be of use and save
608 roundtrips for clients.
609 module-config: <"module names">
611 Module configuration, a list of module names separated by spa‐
612 ces, surround the string with quotes (""). The modules can be
613 validator, iterator. Setting this to "iterator" will result in
614 a non-validating server. Setting this to "validator iterator"
615 will turn on DNSSEC validation. The ordering of the modules is
616 important. You must also set trust-anchors for validation to be
617 useful.
618 trust-anchor-file: <filename>
620 File with trusted keys for validation. Both DS and DNSKEY
621 entries can appear in the file. The format of the file is the
622 standard DNS Zone file format. Default is "", or no trust
623 anchor file.
624 auto-trust-anchor-file: <filename>
626 File with trust anchor for one zone, which is tracked with
627 RFC5011 probes. The probes are several times per month, thus
628 the machine must be online frequently. The initial file can be
629 one with contents as described in trust-anchor-file. The file
630 is written to when the anchor is updated, so the unbound user
631 must have write permission.
632 trust-anchor: <"Resource Record">
634 A DS or DNSKEY RR for a key to use for validation. Multiple
635 entries can be given to specify multiple trusted keys, in addi‐
636 tion to the trust-anchor-files. The resource record is entered
637 in the same format as 'dig' or 'drill' prints them, the same
638 format as in the zone file. Has to be on a single line, with ""
639 around it. A TTL can be specified for ease of cut and paste, but
640 is ignored. A class can be specified, but class IN is default.
641 trusted-keys-file: <filename>
643 File with trusted keys for validation. Specify more than one
644 file with several entries, one file per entry. Like
645 trust-anchor-file but has a different file format. Format is
646 BIND-9 style format, the trusted-keys { name flag proto algo
647 "key"; }; clauses are read. It is possible to use wildcards
648 with this statement, the wildcard is expanded on start and on
649 reload.
650 dlv-anchor-file: <filename>
652 File with trusted keys for DLV (DNSSEC Lookaside Validation).
653 Both DS and DNSKEY entries can be used in the file, in the same
654 format as for trust-anchor-file: statements. Only one DLV can be
655 configured, more would be slow. The DLV configured is used as a
656 root trusted DLV, this means that it is a lookaside for the
657 root. Default is "", or no dlv anchor file.
658 dlv-anchor: <"Resource Record">
660 Much like trust-anchor, this is a DLV anchor with the DS or
661 DNSKEY inline.
662 domain-insecure: <domain name>
664 Sets domain name to be insecure, DNSSEC chain of trust is
665 ignored towards the domain name. So a trust anchor above the
666 domain name can not make the domain secure with a DS record,
667 such a DS record is then ignored. Also keys from DLV are
668 ignored for the domain. Can be given multiple times to specify
669 multiple domains that are treated as if unsigned. If you set
670 trust anchors for the domain they override this setting (and the
671 domain is secured).
672 This can be useful if you want to make sure a trust anchor for
674 external lookups does not affect an (unsigned) internal domain.
675 A DS record externally can create validation failures for that
676 internal domain.
677 val-override-date: <rrsig-style date spec>
679 Default is "" or "0", which disables this debugging feature. If
680 enabled by giving a RRSIG style date, that date is used for ver‐
681 ifying RRSIG inception and expiration dates, instead of the cur‐
682 rent date. Do not set this unless you are debugging signature
683 inception and expiration. The value -1 ignores the date alto‐
684 gether, useful for some special applications.
685 val-sig-skew-min: <seconds>
687 Minimum number of seconds of clock skew to apply to validated
688 signatures. A value of 10% of the signature lifetime (expira‐
689 tion - inception) is used, capped by this setting. Default is
690 3600 (1 hour) which allows for daylight savings differences.
691 Lower this value for more strict checking of short lived signa‐
692 tures.
693 val-sig-skew-max: <seconds>
695 Maximum number of seconds of clock skew to apply to validated
696 signatures. A value of 10% of the signature lifetime (expira‐
697 tion - inception) is used, capped by this setting. Default is
698 86400 (24 hours) which allows for timezone setting problems in
699 stable domains. Setting both min and max very low disables the
700 clock skew allowances. Setting both min and max very high makes
701 the validator check the signature timestamps less strictly.
702 val-bogus-ttl: <number>
704 The time to live for bogus data. This is data that has failed
705 validation; due to invalid signatures or other checks. The TTL
706 from that data cannot be trusted, and this value is used
707 instead. The value is in seconds, default 60. The time interval
708 prevents repeated revalidation of bogus data.
709 val-clean-additional: <yes or no>
711 Instruct the validator to remove data from the additional sec‐
712 tion of secure messages that are not signed properly. Messages
713 that are insecure, bogus, indeterminate or unchecked are not
714 affected. Default is yes. Use this setting to protect the users
715 that rely on this validator for authentication from protentially
716 bad data in the additional section.
717 val-log-level: <number>
719 Have the validator print validation failures to the log.
720 Regardless of the verbosity setting. Default is 0, off. At 1,
721 for every user query that fails a line is printed to the logs.
722 This way you can monitor what happens with validation. Use a
723 diagnosis tool, such as dig or drill, to find out why validation
724 is failing for these queries. At 2, not only the query that
725 failed is printed but also the reason why unbound thought it was
726 wrong and which server sent the faulty data.
727 val-permissive-mode: <yes or no>
729 Instruct the validator to mark bogus messages as indeterminate.
730 The security checks are performed, but if the result is bogus
731 (failed security), the reply is not withheld from the client
732 with SERVFAIL as usual. The client receives the bogus data. For
733 messages that are found to be secure the AD bit is set in
734 replies. Also logging is performed as for full validation. The
735 default value is "no".
736 ignore-cd-flag: <yes or no>
738 Instruct unbound to ignore the CD flag from clients and refuse
739 to return bogus answers to them. Thus, the CD (Checking Dis‐
740 abled) flag does not disable checking any more. This is useful
741 if legacy (w2008) servers that set the CD flag but cannot vali‐
742 date DNSSEC themselves are the clients, and then unbound pro‐
743 vides them with DNSSEC protection. The default value is "no".
744 val-nsec3-keysize-iterations: <"list of values">
746 List of keysize and iteration count values, separated by spaces,
747 surrounded by quotes. Default is "1024 150 2048 500 4096 2500".
748 This determines the maximum allowed NSEC3 iteration count before
749 a message is simply marked insecure instead of performing the
750 many hashing iterations. The list must be in ascending order and
751 have at least one entry. If you set it to "1024 65535" there is
752 no restriction to NSEC3 iteration values. This table must be
753 kept short; a very long list could cause slower operation.
754 add-holddown: <seconds>
756 Instruct the auto-trust-anchor-file probe mechanism for RFC5011
757 autotrust updates to add new trust anchors only after they have
758 been visible for this time. Default is 30 days as per the RFC.
759 del-holddown: <seconds>
761 Instruct the auto-trust-anchor-file probe mechanism for RFC5011
762 autotrust updates to remove revoked trust anchors after they
763 have been kept in the revoked list for this long. Default is 30
764 days as per the RFC.
765 keep-missing: <seconds>
767 Instruct the auto-trust-anchor-file probe mechanism for RFC5011
768 autotrust updates to remove missing trust anchors after they
769 have been unseen for this long. This cleans up the state file
770 if the target zone does not perform trust anchor revocation, so
771 this makes the auto probe mechanism work with zones that perform
772 regular (non-5011) rollovers. The default is 366 days. The
773 value 0 does not remove missing anchors, as per the RFC.
774 key-cache-size: <number>
776 Number of bytes size of the key cache. Default is 4 megabytes.
777 A plain number is in bytes, append 'k', 'm' or 'g' for kilo‐
778 bytes, megabytes or gigabytes (1024*1024 bytes in a megabyte).
779 key-cache-slabs: <number>
781 Number of slabs in the key cache. Slabs reduce lock contention
782 by threads. Must be set to a power of 2. Setting (close) to the
783 number of cpus is a reasonable guess.
784 neg-cache-size: <number>
786 Number of bytes size of the aggressive negative cache. Default
787 is 1 megabyte. A plain number is in bytes, append 'k', 'm' or
788 'g' for kilobytes, megabytes or gigabytes (1024*1024 bytes in a
789 megabyte).
790 local-zone: <zone> <type>
792 Configure a local zone. The type determines the answer to give
793 if there is no match from local-data. The types are deny,
794 refuse, static, transparent, redirect, nodefault, typetranspar‐
795 ent, and are explained below. After that the default settings
796 are listed. Use local-data: to enter data into the local zone.
797 Answers for local zones are authoritative DNS answers. By
798 default the zones are class IN.
799 If you need more complicated authoritative data, with referrals,
801 wildcards, CNAME/DNAME support, or DNSSEC authoritative service,
802 setup a stub-zone for it as detailed in the stub zone section
803 below.
804 deny Do not send an answer, drop the query. If there is a match
806 from local data, the query is answered.
807 refuse
809 Send an error message reply, with rcode REFUSED. If there is
810 a match from local data, the query is answered.
811 static
813 If there is a match from local data, the query is answered.
814 Otherwise, the query is answered with nodata or nxdomain.
815 For a negative answer a SOA is included in the answer if
816 present as local-data for the zone apex domain.
817 transparent
819 If there is a match from local data, the query is answered.
820 Otherwise if the query has a different name, the query is
821 resolved normally. If the query is for a name given in
822 localdata but no such type of data is given in localdata,
823 then a noerror nodata answer is returned. If no local-zone
824 is given local-data causes a transparent zone to be created
825 by default.
826 typetransparent
828 If there is a match from local data, the query is answered.
829 If the query is for a different name, or for the same name
830 but for a different type, the query is resolved normally.
831 So, similar to transparent but types that are not listed in
832 local data are resolved normally, so if an A record is in the
833 local data that does not cause a nodata reply for AAAA
834 queries.
835 redirect
837 The query is answered from the local data for the zone name.
838 There may be no local data beneath the zone name. This
839 answers queries for the zone, and all subdomains of the zone
840 with the local data for the zone. It can be used to redirect
841 a domain to return a different address record to the end
842 user, with local-zone: "example.com." redirect and
843 local-data: "example.com. A 127.0.0.1" queries for www.exam‐
844 ple.com and www.foo.example.com are redirected, so that users
845 with web browsers cannot access sites with suffix exam‐
846 ple.com.
847 nodefault
849 Used to turn off default contents for AS112 zones. The other
850 types also turn off default contents for the zone. The 'node‐
851 fault' option has no other effect than turning off default
852 contents for the given zone.
853 The default zones are localhost, reverse 127.0.0.1 and ::1, and the
855 AS112 zones. The AS112 zones are reverse DNS zones for private use and
856 reserved IP addresses for which the servers on the internet cannot pro‐
857 vide correct answers. They are configured by default to give nxdomain
858 (no reverse information) answers. The defaults can be turned off by
859 specifying your own local-zone of that name, or using the 'nodefault'
860 type. Below is a list of the default zone contents.
861 localhost
863 The IP4 and IP6 localhost information is given. NS and SOA
864 records are provided for completeness and to satisfy some DNS
865 update tools. Default content:
866 local-zone: "localhost." static
867 local-data: "localhost. 10800 IN NS localhost."
868 local-data: "localhost. 10800 IN
869 SOA localhost. nobody.invalid. 1 3600 1200 604800 10800"
870 local-data: "localhost. 10800 IN A 127.0.0.1"
871 local-data: "localhost. 10800 IN AAAA ::1"
872 reverse IPv4 loopback
874 Default content:
875 local-zone: "127.in-addr.arpa." static
876 local-data: "127.in-addr.arpa. 10800 IN NS localhost."
877 local-data: "127.in-addr.arpa. 10800 IN
878 SOA localhost. nobody.invalid. 1 3600 1200 604800 10800"
879 local-data: "1.0.0.127.in-addr.arpa. 10800 IN
880 PTR localhost."
881 reverse IPv6 loopback
883 Default content:
884 local-zone: "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.
885 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa." static
886 local-data: "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.
887 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa. 10800 IN
888 NS localhost."
889 local-data: "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.
890 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa. 10800 IN
891 SOA localhost. nobody.invalid. 1 3600 1200 604800 10800"
892 local-data: "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.
893 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa. 10800 IN
894 PTR localhost."
895 reverse RFC1918 local use zones
897 Reverse data for zones 10.in-addr.arpa, 16.172.in-addr.arpa
898 to 31.172.in-addr.arpa, 168.192.in-addr.arpa. The
899 local-zone: is set static and as local-data: SOA and NS
900 records are provided.
901 reverse RFC3330 IP4 this, link-local, testnet and broadcast
903 Reverse data for zones 0.in-addr.arpa, 254.169.in-addr.arpa,
904 2.0.192.in-addr.arpa (TEST NET 1), 100.51.198.in-addr.arpa
905 (TEST NET 2), 113.0.203.in-addr.arpa (TEST NET 3),
906 255.255.255.255.in-addr.arpa.
907 reverse RFC4291 IP6 unspecified
909 Reverse data for zone
910 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.
911 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa.
912 reverse RFC4193 IPv6 Locally Assigned Local Addresses
914 Reverse data for zone D.F.ip6.arpa.
915 reverse RFC4291 IPv6 Link Local Addresses
917 Reverse data for zones 8.E.F.ip6.arpa to B.E.F.ip6.arpa.
918 reverse IPv6 Example Prefix
920 Reverse data for zone 8.B.D.0.1.0.0.2.ip6.arpa. This zone is
921 used for tutorials and examples. You can remove the block on
922 this zone with:
923 local-zone: 8.B.D.0.1.0.0.2.ip6.arpa. nodefault
924 You can also selectively unblock a part of the zone by making
925 that part transparent with a local-zone statement. This also
926 works with the other default zones.
927 local-data: "<resource record string>"
929 Configure local data, which is served in reply to queries for it.
930 The query has to match exactly unless you configure the local-zone
931 as redirect. If not matched exactly, the local-zone type deter‐
932 mines further processing. If local-data is configured that is not
933 a subdomain of a local-zone, a transparent local-zone is config‐
934 ured. For record types such as TXT, use single quotes, as in
935 local-data: 'example. TXT "text"'.
936 If you need more complicated authoritative data, with referrals,
938 wildcards, CNAME/DNAME support, or DNSSEC authoritative service,
939 setup a stub-zone for it as detailed in the stub zone section
940 below.
941 local-data-ptr: "IPaddr name"
943 Configure local data shorthand for a PTR record with the reversed
944 IPv4 or IPv6 address and the host name. For example "192.0.2.4
945 www.example.com". TTL can be inserted like this: "2001:DB8::4
946 7200 www.example.com"
947 Remote Control Options
949 In the remote-control: clause are the declarations for the remote con‐
950 trol facility. If this is enabled, the unbound-control(8) utility can
951 be used to send commands to the running unbound server. The server
952 uses these clauses to setup SSLv3 / TLSv1 security for the connection.
953 The unbound-control(8) utility also reads the remote-control section
954 for options. To setup the correct self-signed certificates use the
955 unbound-control-setup(8) utility.
956 control-enable: <yes or no>
958 The option is used to enable remote control, default is "no". If
959 turned off, the server does not listen for control commands.
960 control-interface: <ip address>
962 Give IPv4 or IPv6 addresses to listen on for control commands. By
963 default localhost (127.0.0.1 and ::1) is listened to. Use 0.0.0.0
964 and ::0 to listen to all interfaces.
965 control-port: <port number>
967 The port number to listen on for control commands, default is
968 8953. If you change this port number, and permissions have been
969 dropped, a reload is not sufficient to open the port again, you
970 must then restart.
971 server-key-file: <private key file>
973 Path to the server private key, by default unbound_server.key.
974 This file is generated by the unbound-control-setup utility. This
975 file is used by the unbound server, but not by unbound-control.
976 server-cert-file: <certificate file.pem>
978 Path to the server self signed certificate, by default
979 unbound_server.pem. This file is generated by the unbound-con‐
980 trol-setup utility. This file is used by the unbound server, and
981 also by unbound-control.
982 control-key-file: <private key file>
984 Path to the control client private key, by default unbound_con‐
985 trol.key. This file is generated by the unbound-control-setup
986 utility. This file is used by unbound-control.
987 control-cert-file: <certificate file.pem>
989 Path to the control client certificate, by default unbound_con‐
990 trol.pem. This certificate has to be signed with the server cer‐
991 tificate. This file is generated by the unbound-control-setup
992 utility. This file is used by unbound-control.
993 Stub Zone Options
995 There may be multiple stub-zone: clauses. Each with a name: and zero or
996 more hostnames or IP addresses. For the stub zone this list of name‐
997 servers is used. Class IN is assumed. The servers should be authority
998 servers, not recursors; unbound performs the recursive processing
999 itself for stub zones.
1000 The stub zone can be used to configure authoritative data to be used by
1002 the resolver that cannot be accessed using the public internet servers.
1003 This is useful for company-local data or private zones. Setup an
1004 authoritative server on a different host (or different port). Enter a
1005 config entry for unbound with stub-addr: <ip address of host[@port]>.
1006 The unbound resolver can then access the data, without referring to the
1007 public internet for it.
1008 This setup allows DNSSEC signed zones to be served by that authorita‐
1010 tive server, in which case a trusted key entry with the public key can
1011 be put in config, so that unbound can validate the data and set the AD
1012 bit on replies for the private zone (authoritative servers do not set
1013 the AD bit). This setup makes unbound capable of answering queries for
1014 the private zone, and can even set the AD bit ('authentic'), but the AA
1015 ('authoritative') bit is not set on these replies.
1016 name: <domain name>
1018 Name of the stub zone.
1019 stub-host: <domain name>
1021 Name of stub zone nameserver. Is itself resolved before it is
1022 used.
1023 stub-addr: <IP address>
1025 IP address of stub zone nameserver. Can be IP 4 or IP 6. To use
1026 a nondefault port for DNS communication append '@' with the port
1027 number.
1028 stub-prime: <yes or no>
1030 This option is by default off. If enabled it performs NS set
1031 priming, which is similar to root hints, where it starts using
1032 the list of nameservers currently published by the zone. Thus,
1033 if the hint list is slightly outdated, the resolver picks up a
1034 correct list online.
1035 stub-first: <yes or no>
1037 If enabled, a query is attempted without the stub clause if it
1038 fails. The data could not be retrieved and would have caused
1039 SERVFAIL because the servers are unreachable, instead it is
1040 tried without this clause. The default is no.
1041 Forward Zone Options
1043 There may be multiple forward-zone: clauses. Each with a name: and zero
1044 or more hostnames or IP addresses. For the forward zone this list of
1045 nameservers is used to forward the queries to. The servers listed as
1046 forward-host: and forward-addr: have to handle further recursion for
1047 the query. Thus, those servers are not authority servers, but are
1048 (just like unbound is) recursive servers too; unbound does not perform
1049 recursion itself for the forward zone, it lets the remote server do it.
1050 Class IN is assumed. A forward-zone entry with name "." and a for‐
1051 ward-addr target will forward all queries to that other server (unless
1052 it can answer from the cache).
1053 name: <domain name>
1055 Name of the forward zone.
1056 forward-host: <domain name>
1058 Name of server to forward to. Is itself resolved before it is
1059 used.
1060 forward-addr: <IP address>
1062 IP address of server to forward to. Can be IP 4 or IP 6. To use
1063 a nondefault port for DNS communication append '@' with the port
1064 number.
1065 forward-first: <yes or no>
1067 If enabled, a query is attempted without the forward clause if
1068 it fails. The data could not be retrieved and would have caused
1069 SERVFAIL because the servers are unreachable, instead it is
1070 tried without this clause. The default is no.
1071 Python Module Options
1073 The python: clause gives the settings for the python(1) script module.
1074 This module acts like the iterator and validator modules do, on queries
1075 and answers. To enable the script module it has to be compiled into
1076 the daemon, and the word "python" has to be put in the module-config:
1077 option (usually first, or between the validator and iterator).
1078 python-script: <python file>
1080 The script file to load.
1081
1083 In the example config settings below memory usage is reduced. Some ser‐
1084 vice levels are lower, notable very large data and a high TCP load are
1085 no longer supported. Very large data and high TCP loads are exceptional
1086 for the DNS. DNSSEC validation is enabled, just add trust anchors. If
1087 you do not have to worry about programs using more than 3 Mb of memory,
1088 the below example is not for you. Use the defaults to receive full ser‐
1089 vice, which on BSD-32bit tops out at 30-40 Mb after heavy usage.
1090 # example settings that reduce memory usage
1092 server:
1093 num-threads: 1
1094 outgoing-num-tcp: 1 # this limits TCP service, uses less buffers.
1095 incoming-num-tcp: 1
1096 outgoing-range: 60 # uses less memory, but less performance.
1097 msg-buffer-size: 8192 # note this limits service, 'no huge stuff'.
1098 msg-cache-size: 100k
1099 msg-cache-slabs: 1
1100 rrset-cache-size: 100k
1101 rrset-cache-slabs: 1
1102 infra-cache-numhosts: 200
1103 infra-cache-slabs: 1
1104 key-cache-size: 100k
1105 key-cache-slabs: 1
1106 neg-cache-size: 10k
1107 num-queries-per-thread: 30
1108 target-fetch-policy: "2 1 0 0 0 0"
1109 harden-large-queries: "yes"
1110 harden-short-bufsize: "yes"
1111
1113 /etc/unbound
1114 default unbound working directory.
1115 /etc/unbound
1117 default chroot(2) location.
1118 /etc/unbound/unbound.conf
1120 unbound configuration file.
1121 /var/run/unbound/unbound.pid
1123 default unbound pidfile with process ID of the running daemon.
1124 unbound.log
1126 unbound log file. default is to log to syslog(3).
1127
1129 unbound(8), unbound-checkconf(8).
1130
1132 Unbound was written by NLnet Labs. Please see CREDITS file in the dis‐
1133 tribution for further details.
1134NLnet Labs Mar 21, 2013 unbound.conf(5)