NAME¶
systemd-resolved.service, systemd-resolved - Network Name
Resolution manager
SYNOPSIS¶
systemd-resolved.service
/lib/systemd/systemd-resolved
DESCRIPTION¶
systemd-resolved is a system service that provides network
name resolution to local applications. It implements a caching and
validating DNS/DNSSEC stub resolver, as well as an LLMNR and MulticastDNS
resolver and responder. Local applications may submit network name
resolution requests via three interfaces:
•The native, fully-featured API
systemd-resolved exposes on the bus, see
org.freedesktop.resolve1(5) and
org.freedesktop.LogControl1(5)
for details. Usage of this API is generally recommended to clients as it is
asynchronous and fully featured (for example, properly returns DNSSEC
validation status and interface scope for addresses as necessary for
supporting link-local networking).
•The glibc
getaddrinfo(3) API as defined by
RFC3493[1] and its related resolver functions, including
gethostbyname(3). This API is widely supported, including beyond the
Linux platform. In its current form it does not expose DNSSEC validation
status information however, and is synchronous only. This API is backed by the
glibc Name Service Switch (
nss(5)). Usage of the glibc NSS module
nss-resolve(8) is required in order to allow glibc's NSS resolver
functions to resolve hostnames via
systemd-resolved.
•Additionally,
systemd-resolved provides a
local DNS stub listener on the IP addresses 127.0.0.53 and 127.0.0.54 on the
local loopback interface. Programs issuing DNS requests directly, bypassing
any local API may be directed to this stub, in order to connect them to
systemd-resolved. Note however that it is strongly recommended that
local programs use the glibc NSS or bus APIs instead (as described above), as
various network resolution concepts (such as link-local addressing, or LLMNR
Unicode domains) cannot be mapped to the unicast DNS protocol.
The DNS stub resolver on 127.0.0.53 provides the full feature set
of the local resolver, which includes offering LLMNR/MulticastDNS
resolution. The DNS stub resolver on 127.0.0.54 provides a more limited
resolver, that operates in "proxy" mode only, i.e. it will pass
most DNS messages relatively unmodified to the current upstream DNS servers
and back, but not try to process the messages locally, and hence does not
validate DNSSEC, or offer up LLMNR/MulticastDNS. (It will translate to
DNS-over-TLS communication if needed however.)
The DNS servers contacted are determined from the global settings
in /etc/systemd/resolved.conf, the per-link static settings in
/etc/systemd/network/*.network files (in case
systemd-networkd.service(8) is used), the per-link dynamic settings
received over DHCP, information provided via resolvectl(1), and any
DNS server information made available by other system services. See
resolved.conf(5) and systemd.network(5) for details about
systemd's own configuration files for DNS servers. To improve compatibility,
/etc/resolv.conf is read in order to discover configured system DNS servers,
but only if it is not a symlink to /run/systemd/resolve/stub-resolv.conf,
/usr/lib/systemd/resolv.conf or /run/systemd/resolve/resolv.conf (see
below).
SYNTHETIC RECORDS¶
systemd-resolved synthesizes DNS resource records (RRs) for
the following cases:
•The local, configured hostname is resolved to all
locally configured IP addresses ordered by their scope, or — if none
are configured — the IPv4 address 127.0.0.2 (which is on the local
loopback interface) and the IPv6 address ::1 (which is the local host).
•The hostnames "localhost" and
"localhost.localdomain" as well as any hostname ending in
".localhost" or ".localhost.localdomain" are resolved to
the IP addresses 127.0.0.1 and ::1.
•The hostname "_gateway" is resolved to
all current default routing gateway addresses, ordered by their metric. This
assigns a stable hostname to the current gateway, useful for referencing it
independently of the current network configuration state.
•The hostname "_outbound" is resolved to
the local IPv4 and IPv6 addresses that are most likely used for communication
with other hosts. This is determined by requesting a routing decision to the
configured default gateways from the kernel and then using the local IP
addresses selected by this decision. This hostname is only available if there
is at least one local default gateway configured. This assigns a stable
hostname to the local outbound IP addresses, useful for referencing them
independently of the current network configuration state.
•The mappings defined in /etc/hosts are resolved
to their configured addresses and back, but they will not affect lookups for
non-address types (like MX). Support for /etc/hosts may be disabled with
ReadEtcHosts=no, see
resolved.conf(5).
PROTOCOLS AND ROUTING¶
The lookup requests that systemd-resolved.service receives are
routed to the available DNS servers, LLMNR, and MulticastDNS interfaces
according to the following rules:
•Names for which synthetic records are generated
(the local hostname, "localhost" and "localdomain", local
gateway, as listed in the previous section) and addresses configured in
/etc/hosts are never routed to the network and a reply is sent
immediately.
•Single-label names are resolved using LLMNR on
all local interfaces where LLMNR is enabled. Lookups for IPv4 addresses are
only sent via LLMNR on IPv4, and lookups for IPv6 addresses are only sent via
LLMNR on IPv6. Note that lookups for single-label synthesized names are not
routed to LLMNR, MulticastDNS or unicast DNS.
•Queries for the address records (A and AAAA) of
single-label non-synthesized names are resolved via unicast DNS using search
domains. For any interface which defines search domains, such look-ups are
routed to the servers defined for that interface, suffixed with each of those
search domains. When global search domains are defined, such look-ups are
routed to the global servers. For each search domain, queries are performed by
suffixing the name with each of the search domains in turn. Additionally,
lookup of single-label names via unicast DNS may be enabled with the
ResolveUnicastSingleLabel=yes setting. The details of which servers are
queried and how the final reply is chosen are described below. Note that this
means that address queries for single-label names are never sent out to remote
DNS servers by default, and resolution is only possible if search domains are
defined.
•Multi-label names with the domain suffix
".local" are resolved using MulticastDNS on all local interfaces
where MulticastDNS is enabled. As with LLMNR, IPv4 address lookups are sent
via IPv4 and IPv6 address lookups are sent via IPv6.
•Queries for multi-label names are routed via
unicast DNS on local interfaces that have a DNS server configured, plus the
globally configured DNS servers if there are any. Which interfaces are used is
determined by the routing logic based on search and route-only domains,
described below. Note that by default, lookups for domains with the
".local" suffix are not routed to DNS servers, unless the domain is
specified explicitly as routing or search domain for the DNS server and
interface. This means that on networks where the ".local" domain is
defined in a site-specific DNS server, explicit search or routing domains need
to be configured to make lookups work within this DNS domain. Note that these
days, it's generally recommended to avoid defining ".local" in a DNS
server, as RFC6762[2] reserves this domain for exclusive MulticastDNS
use.
•Address lookups (reverse lookups) are routed
similarly to multi-label names, with the exception that addresses from the
link-local address range are never routed to unicast DNS and are only resolved
using LLMNR and MulticastDNS (when enabled).
If lookups are routed to multiple interfaces, the first successful
response is returned (thus effectively merging the lookup zones on all
matching interfaces). If the lookup failed on all interfaces, the last
failing response is returned.
Routing of lookups is determined by the per-interface routing
domains (search and route-only) and global search domains. See
systemd.network(5) and resolvectl(1) for a description how
those settings are set dynamically and the discussion of Domains= in
resolved.conf(5) for a description of globally configured DNS
settings.
The following query routing logic applies for unicast DNS lookups
initiated by systemd-resolved.service:
•If a name to look up matches (that is: is equal
to or has as suffix) any of the configured routing domains (search or
route-only) of any link, or the globally configured DNS settings, "best
matching" routing domain is determined: the matching one with the most
labels. The query is then sent to all DNS servers of any links or the globally
configured DNS servers associated with this "best matching" routing
domain. (Note that more than one link might have this same "best
matching" routing domain configured, in which case the query is sent to
all of them in parallel).
In case of single-label names, when search domains are defined,
the same logic applies, except that the name is first suffixed by each of
the search domains in turn. Note that this search logic doesn't apply to any
names with at least one dot. Also see the discussion about compatibility
with the traditional glibc resolver below.
•If a query does not match any configured routing
domain (either per-link or global), it is sent to all DNS servers that are
configured on links with the DefaultRoute= option set, as well as the
globally configured DNS server.
•If there is no link configured as
DefaultRoute= and no global DNS server configured, one of the
compiled-in fallback DNS servers is used.
•Otherwise the unicast DNS query fails, as no
suitable DNS servers can be determined.
The DefaultRoute= option is a boolean setting configurable
with resolvectl or in .network files. If not set, it is implicitly
determined based on the configured DNS domains for a link: if there's a
route-only domain other than "~.", it defaults to false, otherwise
to true.
Effectively this means: in order to support single-label
non-synthesized names, define appropriate search domains. In order to
preferably route all DNS queries not explicitly matched by routing domain
configuration to a specific link, configure a "~." route-only
domain on it. This will ensure that other links will not be considered for
these queries (unless they too carry such a routing domain). In order to
route all such DNS queries to a specific link only if no other link is
preferred, set the DefaultRoute= option for the link to true and do
not configure a "~." route-only domain on it. Finally, in order to
ensure that a specific link never receives any DNS traffic not matching any
of its configured routing domains, set the DefaultRoute= option for
it to false.
See org.freedesktop.resolve1(5) for information about the
D-Bus APIs systemd-resolved provides.
COMPATIBILITY WITH THE TRADITIONAL GLIBC STUB RESOLVER¶
This section provides a short summary of differences in the
resolver implemented by nss-resolve(8) together with
systemd-resolved and the traditional stub resolver implemented in
nss-dns.
•Some names are always resolved internally (see
Synthetic Records above). Traditionally they would be resolved by nss-files if
provided in /etc/hosts. But note that the details of how a query is
constructed are under the control of the client library. nss-dns will first
try to resolve names using search domains and even if those queries are routed
to systemd-resolved, it will send them out over the network using the usual
rules for multi-label name routing [3].
•Single-label names are not resolved for A and
AAAA records using unicast DNS (unless overridden with
ResolveUnicastSingleLabel=, see
resolved.conf(5)). This is
similar to the
no-tld-query option being set in
resolv.conf(5).
•Search domains are not used for suffixing
of multi-label names. (Search domains are nevertheless used for lookup
routing, for names that were originally specified as single-label or
multi-label.) Any name with at least one dot is always interpreted as a FQDN.
nss-dns would resolve names both as relative (using search domains) and
absolute FQDN names. Some names would be resolved as relative first, and after
that query has failed, as absolute, while other names would be resolved in
opposite order. The ndots option in /etc/resolv.conf was used to
control how many dots the name needs to have to be resolved as relative first.
This stub resolver does not implement this at all: multi-label names are only
resolved as FQDNs.[4]
•This resolver has a notion of the special
".local" domain used for MulticastDNS, and will not route queries
with that suffix to unicast DNS servers unless explicitly configured, see
above. Also, reverse lookups for link-local addresses are not sent to unicast
DNS servers.
•This resolver reads and caches /etc/hosts
internally. (In other words, nss-resolve replaces nss-files in addition to
nss-dns). Entries in /etc/hosts have highest priority.
•This resolver also implements LLMNR and
MulticastDNS in addition to the classic unicast DNS protocol, and will resolve
single-label names using LLMNR (when enabled) and names ending in
".local" using MulticastDNS (when enabled).
•Environment variables
$LOCALDOMAIN and
$RES_OPTIONS described in
resolv.conf(5) are not supported
currently.
•The nss-dns resolver maintains little state
between subsequent DNS queries, and for each query always talks to the first
listed DNS server from /etc/resolv.conf first, and on failure continues with
the next until reaching the end of the list which is when the query fails. The
resolver in systemd-resolved.service however maintains state, and will
continuously talk to the same server for all queries on a particular lookup
scope until some form of error is seen at which point it switches to the next,
and then continuously stays with it for all queries on the scope until the
next failure, and so on, eventually returning to the first configured server.
This is done to optimize lookup times, in particular given that the resolver
typically must first probe server feature sets when talking to a server, which
is time consuming. This different behaviour implies that listed DNS servers
per lookup scope must be equivalent in the zones they serve, so that sending a
query to one of them will yield the same results as sending it to another
configured DNS server.
/ETC/RESOLV.CONF¶
Four modes of handling /etc/resolv.conf (see
resolv.conf(5)) are supported:
•systemd-resolved maintains the
/run/systemd/resolve/stub-resolv.conf file for compatibility with traditional
Linux programs. This file lists the 127.0.0.53 DNS stub (see above) as the
only DNS server. It also contains a list of search domains that are in use by
systemd-resolved. The list of search domains is always kept up-to-date. Note
that /run/systemd/resolve/stub-resolv.conf should not be used directly by
applications, but only through a symlink from /etc/resolv.conf. This file may
be symlinked from /etc/resolv.conf in order to connect all local clients that
bypass local DNS APIs to systemd-resolved with correct search domains
settings. This mode of operation is recommended.
•A static file /usr/lib/systemd/resolv.conf is
provided that lists the 127.0.0.53 DNS stub (see above) as only DNS server.
This file may be symlinked from /etc/resolv.conf in order to connect all local
clients that bypass local DNS APIs to systemd-resolved. This file does
not contain any search domains.
•systemd-resolved maintains the
/run/systemd/resolve/resolv.conf file for compatibility with traditional Linux
programs. This file may be symlinked from /etc/resolv.conf and is always kept
up-to-date, containing information about all known DNS servers. Note the file
format's limitations: it does not know a concept of per-interface DNS servers
and hence only contains system-wide DNS server definitions. Note that
/run/systemd/resolve/resolv.conf should not be used directly by applications,
but only through a symlink from /etc/resolv.conf. If this mode of operation is
used local clients that bypass any local DNS API will also bypass
systemd-resolved and will talk directly to the known DNS servers.
•Alternatively, /etc/resolv.conf may be managed by
other packages, in which case systemd-resolved will read it for DNS
configuration data. In this mode of operation systemd-resolved is
consumer rather than provider of this configuration file.
Note that the selected mode of operation for this file is detected
fully automatically, depending on whether /etc/resolv.conf is a symlink to
/run/systemd/resolve/resolv.conf or lists 127.0.0.53 as DNS server.
SIGNALS¶
SIGUSR1
Upon reception of the SIGUSR1 process signal
systemd-resolved will dump the contents of all DNS resource record
caches it maintains, as well as all feature level information it learnt about
configured DNS servers into the system logs.
SIGUSR2
Upon reception of the SIGUSR2 process signal
systemd-resolved will flush all caches it maintains. Note that it
should normally not be necessary to request this explicitly – except
for debugging purposes – as systemd-resolved flushes the caches
automatically anyway any time the host's network configuration changes.
Sending this signal to systemd-resolved is equivalent to the
resolvectl flush-caches command, however the latter is recommended
since it operates in a synchronous way.
SIGRTMIN+1
Upon reception of the SIGRTMIN+1 process signal
systemd-resolved will forget everything it learnt about the configured
DNS servers. Specifically any information about server feature support is
flushed out, and the server feature probing logic is restarted on the next
request, starting with the most fully featured level. Note that it should
normally not be necessary to request this explicitly – except for
debugging purposes – as systemd-resolved automatically forgets
learnt information any time the DNS server configuration changes. Sending this
signal to systemd-resolved is equivalent to the resolvectl
reset-server-features command, however the latter is recommended since it
operates in a synchronous way.
NOTES¶
- 1.
- RFC3493
- 2.
- RFC6762
- 3.
- For example, if /etc/resolv.conf has
nameserver 127.0.0.53
search foobar.com barbar.com
and we look up "localhost", nss-dns will send the
following queries to systemd-resolved listening on 127.0.0.53:53: first
"localhost.foobar.com", then "localhost.barbar.com", and
finally "localhost". If (hopefully) the first two queries fail,
systemd-resolved will synthesize an answer for the third query.
When using nss-dns with any search domains, it is thus
crucial to always configure nss-files with higher priority and provide
mappings for names that should not be resolved using search domains.
- 4.
- There are currently more than 1500 top-level domain names defined, and new
ones are added regularly, often using "attractive" names that
are also likely to be used locally. Not looking up multi-label names in
this fashion avoids fragility in both directions: a valid global name
could be obscured by a local name, and resolution of a relative local name
could suddenly break when a new top-level domain is created, or when a new
subdomain of a top-level domain in registered. Resolving any given name as
either relative or absolute avoids this ambiguity.
