]> git.ipfire.org Git - thirdparty/systemd.git/blob - docs/USER_GROUP_API.md
Merge pull request #15697 from OhNoMoreGit/fix-path-units
[thirdparty/systemd.git] / docs / USER_GROUP_API.md
1 ---
2 title: User/Group Record Lookup API via Varlink
3 category: Users, Groups and Home Directories
4 layout: default
5 ---
6
7 # User/Group Record Lookup API via Varlink
8
9 JSON User/Group Records (as described in the [JSON User
10 Records](https://systemd.io/USER_RECORD) and [JSON Group
11 Records](https://systemd.io/GROUP_RECORD) documents) that are defined on the
12 local system may be queried with a [Varlink](https://varlink.org/) API. This
13 API takes both the role of what
14 [`getpwnam(3)`](http://man7.org/linux/man-pages/man3/getpwnam.3.html) and
15 related calls are for `struct passwd`, as well as the interfaces modules
16 implementing the [glibc Name Service Switch
17 (NSS)](https://www.gnu.org/software/libc/manual/html_node/Name-Service-Switch.html)
18 expose. Or in other words, it both allows applications to efficiently query
19 user/group records from local services, and allows local subsystems to provide
20 user/group records efficiently to local applications.
21
22 This simple API only exposes only three method calls, and requires only a small
23 subset of the Varlink functionality.
24
25 ## Why Varlink?
26
27 The API described in this document is based on a simple subset of the
28 mechanisms described by [Varlink](https://varlink.org/). The choice of
29 preferring Varlink over D-Bus and other IPCs in this context was made for three
30 reasons:
31
32 1. User/Group record resolution should work during early boot and late shutdown
33 without special handling. This is very hard to do with D-Bus, as the broker
34 service for D-Bus generally runs as regular system daemon and is hence only
35 available at the latest boot stage.
36
37 2. The JSON user/group records are native JSON data, hence picking an IPC
38 system that natively operates with JSON data is natural and clean.
39
40 3. IPC systems such as D-Bus do not provide flow control and are thus unusable
41 for streaming data. They are useful to pass around short control messages,
42 but as soon as potentially many and large objects shall be transferred,
43 D-Bus is not suitable, as any such streaming of messages would be considered
44 flooding in D-Bus' logic, and thus possibly result in termination of
45 communication. Since the APIs defined in this document need to support
46 enumerating potentially large numbers of users and groups, D-Bus is simply
47 not an appropriate option.
48
49 ## Concepts
50
51 Each subsystem that needs to define users and groups on the local system is
52 supposed to implement this API, and offer its interfaces on a Varlink
53 `AF_UNIX`/`SOCK_STREAM` file system socket bound into the
54 `/run/systemd/userdb/` directory. When a client wants to look up a user or
55 group record, it contacts all sockets bound in this directory in parallel, and
56 enqueues the same query to each. The first positive reply is then returned to
57 the application, or if all fail the last seen error is returned
58 instead. (Alternatively a special Varlink service is available,
59 `io.systemd.Multiplexer` which acts as frontend and will do the parallel
60 queries on behalf of the client, drastically simplifying client
61 development. This service is not available during earliest boot and final
62 shutdown phases.)
63
64 Unlike with glibc NSS there's no order or programmatic expression language
65 defined in which queries are issued to the various services. Instead, all
66 queries are always enqueued in parallel to all defined services, in order to
67 make look-ups efficient, and the simple rule of "first successful lookup wins"
68 is unconditionally followed for user and group look-ups (though not for
69 membership lookups, see below).
70
71 This simple scheme only works safely as long as every service providing
72 user/group records carefully makes sure not to answer with conflicting
73 records. This API does not define any mechanisms for dealing with user/group
74 name/ID collisions during look-up nor during record registration. It assumes
75 the various subsystems that want to offer user and group records to the rest of
76 the system have made sufficiently sure in advance that their definitions do not
77 collide with those of other services. Clients are not expected to merge
78 multiple definitions for the same user or group, and will also not be able to
79 detect conflicts and suppress such conflicting records.
80
81 It is recommended to name the sockets in the directory in reverse domain name
82 notation, but this is neither required nor enforced.
83
84 ## Well-Known Services
85
86 Any subsystem that wants to provide user/group records can do so, simply by
87 binding a socket in the aforementioned directory. By default two
88 services are listening there, that have special relevance:
89
90 1. `io.systemd.NameServiceSwitch` → This service makes the classic UNIX/glibc
91 NSS user/group records available as JSON User/Group records. Any such
92 records are automatically converted as needed, and possibly augmented with
93 information from the shadow databases.
94
95 2. `io.systemd.Multiplexer` → This service multiplexes client queries to all
96 other running services. It's supposed to simplify client development: in
97 order to look up or enumerate user/group records it's sufficient to talk to
98 one service instead of all of them in parallel. Note that it is not available
99 during earliest boot and final shutdown phases, hence for programs running
100 in that context it is preferable to implement the parallel lookup
101 themselves.
102
103 Both these services are implemented by the same daemon
104 `systemd-userdbd.service`.
105
106 Note that these services currently implement a subset of Varlink only. For
107 example, introspection is not available, and the resolver logic is not used.
108
109 ## Other Services
110
111 The `systemd` project provides two other services implementing this
112 interface. Specifically:
113
114 1. `io.systemd.DynamicUser` → This service is implemented by the service
115 manager itself, and provides records for the users and groups synthesized
116 via `DynamicUser=` in unit files.
117
118 2. `io.systemd.Home` → This service is implemented by `systemd-homed.service`
119 and provides records for the users and groups defined by the home
120 directories it manages.
121
122 Other projects are invited to implement these services too. For example it
123 would make sense for LDAP/ActiveDirectory projects to implement these
124 interfaces, which would provide them a way to do per-user resource management
125 enforced by systemd and defined directly in LDAP directories.
126
127 ## Compatibility with NSS
128
129 Two-way compatibility with classic UNIX/glibc NSS user/group records is
130 provided. When using the Varlink API, lookups into databases provided only via
131 NSS (and not natively via Varlink) are handled by the
132 `io.systemd.NameServiceSwitch` service (see above). When using the NSS API
133 (i.e. `getpwnam()` and friends) the `nss-systemd` module will automatically
134 synthesize NSS records for users/groups natively defined via a Varlink
135 API. Special care is taken to avoid recursion between these two compatibility
136 mechanisms.
137
138 Subsystems that shall provide user/group records to the system may choose
139 between offering them via an NSS module or via a this Varlink API, either way
140 all records are accessible via both APIs, due to the bidirectional
141 forwarding. It is also possible to provide the same records via both APIs
142 directly, but in that case the compatibility logic must be turned off. There
143 are mechanisms in place for this, please contact the systemd project for
144 details, as these are currently not documented.
145
146 ## Caching of User Records
147
148 This API defines no concepts for caching records. If caching is desired it
149 should be implemented in the subsystems that provide the user records, not in
150 the clients consuming them.
151
152 ## Method Calls
153
154 ```
155 interface io.systemd.UserDatabase
156
157 method GetUserRecord(
158 uid : ?int,
159 userName : ?string,
160 service : string
161 ) -> (
162 record : object,
163 incomplete : bool
164 )
165
166 method GetGroupRecord(
167 gid : ?int,
168 groupName : ?string,
169 service : string
170 ) -> (
171 record : object,
172 incomplete : bool
173 )
174
175 method GetMemberships(
176 userName : ?string,
177 groupName : ?string,
178 service : string
179 ) -> (
180 userName : string,
181 groupName : string
182 )
183
184 error NoRecordFound()
185 error BadService()
186 error ServiceNotAvailable()
187 error ConflictingRecordFound()
188 ```
189
190 The `GetUserRecord` method looks up or enumerates a user record. If the `uid`
191 parameter is set it specifies the numeric UNIX UID to search for. If the
192 `userName` parameter is set it specifies the name of the user to search
193 for. Typically, only one of the two parameters are set, depending whether a
194 look-up by UID or by name is desired. However, clients may also specify both
195 parameters, in which case a record matching both will be returned, and if only
196 one exists that matches one of the two parameters but not the other an error of
197 `ConflictingRecordFound` is returned. If neither of the two parameters are set
198 the whole user database is enumerated. In this case the method call needs to be
199 made with `more` set, so that multiple method call replies may be generated as
200 effect, each carrying one user record.
201
202 The `service` parameter is mandatory and should be set to the service name
203 being talked to (i.e. to the same name as the `AF_UNIX` socket path, with the
204 `/run/systemd/userdb/` prefix removed). This is useful to allow implementation
205 of multiple services on the same socket (which is used by
206 `systemd-userdbd.service`).
207
208 The method call returns one or more user records, depending which type of query is
209 used (see above). The record is returned in the `record` field. The
210 `incomplete` field indicates whether the record is complete. Services providing
211 user record lookup should only pass the `privileged` section of user records to
212 clients that either match the user the record is about or to sufficiently
213 privileged clients, for all others the section must be removed so that no
214 sensitive data is leaked this way. The `incomplete` parameter should indicate
215 whether the record has been modified like this or not (i.e. it is `true` if a
216 `privileged` section existed in the user record and was removed, and `false` if
217 no `privileged` section existed or one existed but hasn't been removed).
218
219 If no user record matching the specified UID or name is known the error
220 `NoRecordFound` is returned (this is also returned if neither UID nor name are
221 specified, and hence enumeration requested but the subsystem currently has no
222 users defined).
223
224 If a method call with an incorrectly set `service` field is received
225 (i.e. either not set at all, or not to the service's own name) a `BadService`
226 error is generated. Finally, `ServiceNotAvailable` should be returned when the
227 backing subsystem is not operational for some reason and hence no information
228 about existence or non-existence of a record can be returned nor any user
229 record at all. (The `service` field is defined in order to allow implementation
230 of daemons that provide multiple distinct user/group services over the same
231 `AF_UNIX` socket: in order to correctly determine which service a client wants
232 to talk to the client needs to provide the name in each request.)
233
234 The `GetGroupRecord` method call works analogously but for groups.
235
236 The `GetMemberships` method call may be used to inquire about group
237 memberships. The `userName` and `groupName` arguments take what the name
238 suggests. If one of the two is specified all matching memberships are returned,
239 if neither is specified all known memberships of any user and any group are
240 returned. The return value is a pair of user name and group name, where the
241 user is a member of the group. If both arguments are specified the specified
242 membership will be tested for, but no others, and the pair is returned if it is
243 defined. Unless both arguments are specified the method call needs to be made
244 with `more` set, so that multiple replies can be returned (since typically
245 there are are multiple members per group and also multiple groups a user is
246 member of). As with `GetUserRecord` and `GetGroupRecord` the `service`
247 parameter needs to contain the name of the service being talked to, in order to
248 allow implementation of multiple service within the same IPC socket. In case no
249 matching membership is known `NoRecordFound` is returned. The other two errors
250 are also generated in the same cases as for `GetUserRecord` and
251 `GetGroupRecord`.
252
253 Unlike with `GetUserRecord` and `GetGroupRecord` the lists of memberships
254 returned by services are always combined. Thus unlike the other two calls a
255 membership lookup query has to wait for the last simultaneous query to complete
256 before the complete list is acquired.
257
258 Note that only the `GetMemberships` call is authoritative about memberships of
259 users in groups. i.e. it should not be considered sufficient to check the
260 `memberOf` field of user records and the `members` field of group records to
261 acquire the full list of memberships. The full list can only bet determined by
262 `GetMemberships`, and as mentioned requires merging of these lists of all local
263 services. Result of this is that it can be one service that defines a user A,
264 and another service that defines a group B, and a third service that declares
265 that A is a member of B.
266
267 And that's really all there is to it.