Configuration Reference¶
Description¶
Configuration files for Knot DNS use simplified YAML format. Simplified means that not all of the features are supported.
For the description of configuration items, we have to declare a meaning of the following symbols:
- INT – Integer
- STR – Textual string
- HEXSTR – Hexadecimal string (with
0x
prefix) - BOOL – Boolean value (
on
/off
ortrue
/false
) - TIME – Number of seconds, an integer with possible time multiplier suffix
(
s
~ 1,m
~ 60,h
~ 3600 ord
~ 24 * 3600) - SIZE – Number of bytes, an integer with possible size multiplier suffix
(
B
~ 1,K
~ 1024,M
~ 1024^2 orG
~ 1024^3) - BASE64 – Base64 encoded string
- ADDR – IPv4 or IPv6 address
- DNAME – Domain name
- ... – Multi-valued item, order of the values is preserved
- [ ] – Optional value
- | – Choice
There are 8 main sections (server
, key
, acl
, control
,
remote
, template
, zone
and log
) and module sections with the
mod-
prefix. The most of the sections (excluding server
and
control
) are sequences of settings blocks. Each settings block
begins with a unique identifier, which can be used as a reference from other
sections (such identifier must be defined in advance).
A multi-valued item can be specified either as a YAML sequence:
address: [10.0.0.1, 10.0.0.2]
or as more single-valued items each on an extra line:
address: 10.0.0.1
address: 10.0.0.2
If an item value contains spaces or other special characters, it is necessary
to enclose such value within double quotes "
"
.
Comments¶
A comment begins with a #
character and is ignored during processing.
Also each configuration section or sequence block allows to specify permanent
comment using comment
item which is stored in the server beside the
configuration.
Includes¶
Another configuration file or files, matching a pattern, can be included at the top level in the current file. If the path is not absolute, then it is considered to be relative to the current file. The pattern can be an arbitrary string meeting POSIX glob requirements, e.g. dir/*.conf. Matching files are processed in sorted order.
include: STR
Server section¶
General options related to the server.
server:
identity: [STR]
version: [STR]
nsid: [STR|HEXSTR]
rundir: STR
user: STR[:STR]
pidfile: STR
udp-workers: INT
tcp-workers: INT
background-workers: INT
async-start: BOOL
tcp-idle-timeout: TIME
tcp-handshake-timeout: TIME
tcp-reply-timeout: TIME
max-tcp-clients: INT
max-udp-payload: SIZE
rate-limit: INT
rate-limit-slip: INT
rate-limit-table-size: INT
listen: ADDR[@INT] ...
identity¶
An identity of the server returned in the response to the query for TXT
record id.server.
or hostname.bind.
in the CHAOS class (see RFC 4892).
Set empty value to disable.
Default: FQDN hostname
version¶
A version of the server software returned in the response to the query
for TXT record version.server.
or version.bind.
in the CHAOS
class (see RFC 4892). Set empty value to disable.
Default: server version
nsid¶
A DNS name server identifier (see RFC 5001). Set empty value to disable.
Default: FQDN hostname
rundir¶
A path for storing run-time data (PID file, unix sockets, etc.).
Default: ${localstatedir}/run/knot
(configured with --with-rundir=path
)
user¶
A system user with an optional system group (user:group) under which the server is run after starting and binding to interfaces. Linux capabilities are employed if supported.
Default: root:root
udp-workers¶
A number of quering UDP workers (threads).
Default: auto-estimated optimal value based on the number of online CPUs
tcp-workers¶
A number of quering TCP workers (threads).
Default: auto-estimated optimal value based on the number of online CPUs
background-workers¶
A number of workers (threads) used to execute background operations (zone loading, zone updates, etc.).
Default: auto-estimated optimal value based on the number of online CPUs
async-start¶
If enabled, server doesn't wait for the zones to be loaded and starts responding immediately with SERVFAIL answers until the zone loads.
Default: off
tcp-idle-timeout¶
Maximum idle time between requests on a TCP connection. This also limits receiving of a single query, each query must be received in this time limit.
Default: 20
tcp-handshake-timeout¶
Maximum time between newly accepted TCP connection and the first query. This is useful to disconnect inactive connections faster than connections that already made at least 1 meaningful query.
Default: 5
max-tcp-clients¶
A maximum number of TCP clients connected in parallel, set this below the file descriptor limit to avoid resource exhaustion.
Default: 100
rate-limit¶
Rate limiting is based on the token bucket scheme. A rate basically represents a number of tokens available each second. Each response is processed and classified (based on several discriminators, e.g. source netblock, qtype, name, rcode, etc.). Classified responses are then hashed and assigned to a bucket containing number of available tokens, timestamp and metadata. When available tokens are exhausted, response is rejected or enters SLIP (server responds with a truncated response). Number of available tokens is recalculated each second.
Default: 0 (disabled)
rate-limit-table-size¶
Size of the hashtable in a number of buckets. The larger the hashtable, the lesser the probability of a hash collision, but at the expense of additional memory costs. Each bucket is estimated roughly to 32 bytes. The size should be selected as a reasonably large prime due to better hash function distribution properties. Hash table is internally chained and works well up to a fill rate of 90 %, general rule of thumb is to select a prime near 1.2 * maximum_qps.
Default: 393241
rate-limit-slip¶
As attacks using DNS/UDP are usually based on a forged source address, an attacker could deny services to the victim netblock if all responses would be completely blocked. The idea behind SLIP mechanism is to send each Nth response as truncated, thus allowing client to reconnect via TCP for at least some degree of service. It is worth noting, that some responses can't be truncated (e.g. SERVFAIL).
It is advisable not to set the slip interval to a value larger than 2, as too large slip value means more denial of service for legitimate requestors, and introduces excessive timeouts during resolution. On the other hand, slipping truncated answer gives the legitimate requestors a chance to reconnect over TCP.
Default: 1
listen¶
One or more IP addresses where the server listens for incoming queries.
Optional port specification (default is 53) can be appended to each address
using @
separator. Use 0.0.0.0
for all configured IPv4 addresses or
::
for all configured IPv6 addresses.
Default: empty
Key section¶
Shared TSIG keys used to authenticate communication with the server.
key:
- id: DNAME
algorithm: hmac-md5 | hmac-sha1 | hmac-sha224 | hmac-sha256 | hmac-sha384 | hmac-sha512
secret: BASE64
id¶
A key name identifier.
ACL section¶
Access control list rule definition.
acl:
- id: STR
address: ADDR[/INT] ...
key: key_id ...
action: transfer | notify | update | control ...
deny: BOOL
id¶
An ACL rule identifier.
address¶
An ordered list of IP addresses or network subnets. The query must match one of them. Empty value means that address match is not required.
Default: empty
key¶
An ordered list of references to TSIG keys. The query must match one of them. Empty value means that TSIG key is not required.
Default: empty
action¶
An ordered list of allowed actions.
Possible values:
transfer
– Allow zone transfernotify
– Allow incoming notifyupdate
– Allow zone updatescontrol
– Allow remote control
Default: empty
Control section¶
Configuration of the server remote control.
Caution: The control protocol is not encrypted and is susceptible to replay attacks in a short timeframe until message digest expires. For that reason, it is recommended to use default UNIX socket.
control:
listen: ADDR[@INT]
acl: acl_id ...
listen¶
A UNIX socket path or IP address where the server listens for remote control
commands. Optional port specification (default is 5533) can be appended to the
address using @
separator.
Default: rundir/knot.sock
acl¶
An ordered list of references to ACL rules allowing the remote control.
Caution: This option has no effect with UNIX socket.
Default: empty
Remote section¶
Definition of remote servers for zone transfers or notifications.
remote:
- id: STR
address: ADDR[@INT] ...
via: ADDR[@INT] ...
key: key_id
id¶
A remote identifier.
address¶
An ordered list of destination IP addresses which are used for communication
with the remote server. The addresses are tried in sequence unless the
operation is successful. Optional destination port (default is 53)
can be appended to the address using @
separator.
Default: empty
via¶
An ordered list of source IP addresses. The first address with the same family
as the destination address is used. Optional source port (default is random)
can be appended to the address using @
separator.
Default: empty
Template section¶
A template is a shareable zone setting which can be used for configuration of many zones in one place. A special default template (with the default identifier) can be used for global querying configuration or as an implicit configuration if a zone doesn't have another template specified.
template:
- id: STR
global-module: STR/STR ...
# All zone options (excluding 'template' item)
id¶
A template identifier.
global-module¶
An ordered list of references to query modules in the form module_name/module_id. These modules apply to all queries.
Caution: This option is available only for the default template.
Default: empty
Zone section¶
Definition of zones served by the server.
zone:
- domain: DNAME
template: template_id
file: STR
storage: STR
master: remote_id ...
ddns-master: remote_id
notify: remote_id ...
acl: acl_id ...
semantic-checks: BOOL
disable-any: BOOL
zonefile-sync: TIME
ixfr-from-differences: BOOL
max-journal-size: SIZE
dnssec-signing: BOOL
kasp-db: STR
request-edns-option: INT:[HEXSTR]
serial-policy: increment | unixtime
module: STR/STR ...
domain¶
A zone name identifier.
template¶
A reference to a configuration template. If not specified and the default template exists, the default template is used.
Default: empty
file¶
A path to the zone file. Non absolute path is relative to storage. It is also possible to use the following formatters:
%s
– means the current zone name in the textual representation (beware of special characters which are escaped or encoded in the \DDD form where DDD is corresponding decimal ASCII code). The zone name doesn't include the terminating dot, except for the root zone.%%
– means the%
character
Default: storage/%s
.zone
storage¶
A data directory for storing zone files, journal files and timers database.
Default: ${localstatedir}/lib/knot
(configured with --with-storage=path
)
ddns-master¶
A reference to zone primary master server. If not specified, the first master server is used.
Default: empty
notify¶
An ordered list of references to remotes to which notify message is sent if the zone changes.
Default: empty
acl¶
An ordered list of references to ACL rules which can allow or disallow zone transfers, updates or incoming notifies.
Default: empty
semantic-checks¶
If enabled, extra zone file semantic checks are turned on.
Several checks are enabled by default and cannot be turned off. An error in mandatory checks causes zone not to be loaded. An error in extra checks is logged only.
Mandatory checks:
- An extra record together with CNAME record (except for RRSIG and DS)
- CNAME link chain length greater than 10 (including infinite cycles)
- DNAME and CNAME records under the same owner (RFC 2672)
- CNAME and DNAME wildcards pointing to themselves
- SOA record missing in the zone (RFC 1034)
- DNAME records having records under it (DNAME children) (RFC 2672)
Extra checks:
- Missing NS record at the zone apex
- Missing glue A or AAAA records
- Broken or non-cyclic NSEC(3) chain
- Wrong NSEC(3) type bitmap
- Multiple NSEC records at the same node
- Missing NSEC records at authoritative nodes
- Extra record types under the same name as NSEC3 record (this is RFC-valid, but Knot will not serve such a zone correctly)
- NSEC3-unsecured delegation that is not part of Opt-out span
- Wrong original TTL value in NSEC3 records
- Wrong RDATA TTL value in RRSIG record
- Signer name in RRSIG RR not the same as in DNSKEY
- Signed RRSIG
- Not all RRs in the node are signed
- Wrong key flags or wrong key in RRSIG record (not the same as ZSK)
Default: off
disable-any¶
If enabled, all authoritative ANY queries sent over UDP will be answered with an empty response and with the TC bit set. Use this option to minimize the risk of DNS reflection attack.
Default: off
zonefile-sync¶
The time after which the current zone in memory will be synced with a zone file
on the disk (see file). The server will serve the latest
zone even after a restart using zone journal, but the zone file on the disk will
only be synced after zonefile-sync
time has expired (or after manual zone
flush). This is applicable when the zone is updated via IXFR, DDNS or automatic
DNSSEC signing. In order to disable automatic zonefile synchronization, -1 value
can be used (manual zone flush is still possible).
Caution: If you are serving large zones with frequent updates where the immediate sync with a zone file is not desirable, increase the value.
Default: 0 (immediate)
ixfr-from-differences¶
If enabled, the server creates zone differences from changes you made to the zone file upon server reload. This option is relevant only if the server is a master server for the zone.
Caution: This option has no effect with enabled dnssec-signing.
Default: off
dnssec-signing¶
If enabled, automatic DNSSEC signing for the zone is turned on.
Caution: Cannot be enabled on a slave zone.
Default: off
request-edns-option¶
An arbitrary EDNS0 option which is included into a server request (AXFR, IXFR, SOA, or NOTIFY). The value is in the option_code:option_data format.
Default: empty
serial-policy¶
Specifies how the zone serial is updated after a dynamic update or automatic DNSSEC signing. If the serial is changed by the dynamic update, no change is made.
Possible values:
increment
– The serial is incremented according to serial number arithmeticunixtime
– The serial is set to the current unix time
Caution: If your serial was in other than unix time format, be careful with the transition to unix time. It may happen that the new serial will be 'lower' than the old one. If this is the case, the transition should be done by hand (see RFC 1982).
Default: increment
module¶
An ordered list of references to query modules in the form module_name/module_id. These modules apply only to the current zone queries.
Default: empty
Logging section¶
Server can be configured to log to the standard output, standard error output, syslog (or systemd journal if systemd is enabled) or into an arbitrary file.
There are 6 logging severity levels:
critical
– Non-recoverable error resulting in server shutdownerror
– Recoverable error, action should be takenwarning
– Warning that might require user actionnotice
– Server notice or hintinfo
– Informational messagedebug
– Debug messages (must be turned on at compile time)
In the case of missing log section, warning
or more serious messages
will be logged to both standard error output and syslog. The info
and
notice
messages will be logged to standard output.
log:
- target: stdout | stderr | syslog | STR
server: critical | error | warning | notice | info | debug
zone: critical | error | warning | notice | info | debug
any: critical | error | warning | notice | info | debug
target¶
A logging output.
Possible values:
stdout
– Standard outputstderr
– Standard error outputsyslog
– Syslog- file_name – File
server¶
Minimum severity level for messages related to general operation of the server that are logged.
Default: empty
Module dnstap¶
The module dnstap allows query and response logging.
For all queries logging, use this module in the default template. For zone-specific logging, use this module in the proper zone configuration.
mod-dnstap:
- id: STR
sink: STR
id¶
A module identifier.
sink¶
A sink path, which can be either a file or a UNIX socket when prefixed with
unix:
.
Default: empty
Module synth-record¶
This module is able to synthesize either forward or reverse records for the given prefix and subnet.
mod-synth-record:
- id: STR
type: forward | reverse
prefix: STR
origin: DNAME
ttl: INT
network: ADDR[/INT]
id¶
A module identifier.
type¶
The type of generated records.
Possible values:
forward
– Forward recordsreverse
– Reverse records
Default: empty
prefix¶
A record owner prefix.
Caution: prefix doesn’t allow dots, address parts in the synthetic names are separated with a dash.
Default: empty
Module dnsproxy¶
The module catches all unsatisfied queries and forwards them to the indicated server for resolution.
mod-dnsproxy:
- id: STR
remote: remote_id
catch-nxdomain: BOOL
id¶
A module identifier.
catch-nxdomain¶
If enabled, all unsatisfied queries (also applies to local zone lookups) are forwarded.
Default: off