]>
Commit | Line | Data |
---|---|---|
1 | systemd System and Service Manager | |
2 | ||
3 | DETAILS: | |
4 | http://0pointer.de/blog/projects/systemd.html | |
5 | ||
6 | WEB SITE: | |
7 | https://www.freedesktop.org/wiki/Software/systemd | |
8 | ||
9 | GIT: | |
10 | git@github.com:systemd/systemd.git | |
11 | https://github.com/systemd/systemd | |
12 | ||
13 | MAILING LIST: | |
14 | https://lists.freedesktop.org/mailman/listinfo/systemd-devel | |
15 | ||
16 | IRC: | |
17 | #systemd on irc.freenode.org | |
18 | ||
19 | BUG REPORTS: | |
20 | https://github.com/systemd/systemd/issues | |
21 | ||
22 | AUTHOR: | |
23 | Lennart Poettering | |
24 | Kay Sievers | |
25 | ...and many others | |
26 | ||
27 | LICENSE: | |
28 | LGPLv2.1+ for all code | |
29 | - except src/basic/MurmurHash2.c which is Public Domain | |
30 | - except src/basic/siphash24.c which is CC0 Public Domain | |
31 | - except src/journal/lookup3.c which is Public Domain | |
32 | - except src/udev/* which is (currently still) GPLv2, GPLv2+ | |
33 | - except tools/chromiumos/* which is BSD-style | |
34 | ||
35 | REQUIREMENTS: | |
36 | Linux kernel >= 3.13 | |
37 | Linux kernel >= 4.2 for unified cgroup hierarchy support | |
38 | ||
39 | Kernel Config Options: | |
40 | CONFIG_DEVTMPFS | |
41 | CONFIG_CGROUPS (it is OK to disable all controllers) | |
42 | CONFIG_INOTIFY_USER | |
43 | CONFIG_SIGNALFD | |
44 | CONFIG_TIMERFD | |
45 | CONFIG_EPOLL | |
46 | CONFIG_NET | |
47 | CONFIG_SYSFS | |
48 | CONFIG_PROC_FS | |
49 | CONFIG_FHANDLE (libudev, mount and bind mount handling) | |
50 | ||
51 | Kernel crypto/hash API | |
52 | CONFIG_CRYPTO_USER_API_HASH | |
53 | CONFIG_CRYPTO_HMAC | |
54 | CONFIG_CRYPTO_SHA256 | |
55 | ||
56 | udev will fail to work with the legacy sysfs layout: | |
57 | CONFIG_SYSFS_DEPRECATED=n | |
58 | ||
59 | Legacy hotplug slows down the system and confuses udev: | |
60 | CONFIG_UEVENT_HELPER_PATH="" | |
61 | ||
62 | Userspace firmware loading is not supported and should | |
63 | be disabled in the kernel: | |
64 | CONFIG_FW_LOADER_USER_HELPER=n | |
65 | ||
66 | Some udev rules and virtualization detection relies on it: | |
67 | CONFIG_DMIID | |
68 | ||
69 | Support for some SCSI devices serial number retrieval, to | |
70 | create additional symlinks in /dev/disk/ and /dev/tape: | |
71 | CONFIG_BLK_DEV_BSG | |
72 | ||
73 | Required for PrivateNetwork= in service units: | |
74 | CONFIG_NET_NS | |
75 | Note that systemd-localed.service and other systemd units use | |
76 | PrivateNetwork so this is effectively required. | |
77 | ||
78 | Required for PrivateUsers= in service units: | |
79 | CONFIG_USER_NS | |
80 | ||
81 | Optional but strongly recommended: | |
82 | CONFIG_IPV6 | |
83 | CONFIG_AUTOFS4_FS | |
84 | CONFIG_TMPFS_XATTR | |
85 | CONFIG_{TMPFS,EXT4_FS,XFS,BTRFS_FS,...}_POSIX_ACL | |
86 | CONFIG_SECCOMP | |
87 | CONFIG_SECCOMP_FILTER (required for seccomp support) | |
88 | CONFIG_CHECKPOINT_RESTORE (for the kcmp() syscall) | |
89 | ||
90 | Required for CPUShares= in resource control unit settings | |
91 | CONFIG_CGROUP_SCHED | |
92 | CONFIG_FAIR_GROUP_SCHED | |
93 | ||
94 | Required for CPUQuota= in resource control unit settings | |
95 | CONFIG_CFS_BANDWIDTH | |
96 | ||
97 | Required for IPAddressDeny= and IPAddressAllow= in resource control | |
98 | unit settings | |
99 | CONFIG_CGROUP_BPF | |
100 | ||
101 | For UEFI systems: | |
102 | CONFIG_EFIVAR_FS | |
103 | CONFIG_EFI_PARTITION | |
104 | ||
105 | We recommend to turn off Real-Time group scheduling in the | |
106 | kernel when using systemd. RT group scheduling effectively | |
107 | makes RT scheduling unavailable for most userspace, since it | |
108 | requires explicit assignment of RT budgets to each unit whose | |
109 | processes making use of RT. As there's no sensible way to | |
110 | assign these budgets automatically this cannot really be | |
111 | fixed, and it's best to disable group scheduling hence. | |
112 | CONFIG_RT_GROUP_SCHED=n | |
113 | ||
114 | It's a good idea to disable the implicit creation of networking bonding | |
115 | devices by the kernel networking bonding module, so that the | |
116 | automatically created "bond0" interface doesn't conflict with any such | |
117 | device created by systemd-networkd (or other tools). Ideally there | |
118 | would be a kernel compile-time option for this, but there currently | |
119 | isn't. The next best thing is to make this change through a modprobe.d | |
120 | drop-in. This is shipped by default, see modprobe.d/systemd.conf. | |
121 | ||
122 | Required for systemd-nspawn: | |
123 | CONFIG_DEVPTS_MULTIPLE_INSTANCES or Linux kernel >= 4.7 | |
124 | ||
125 | Note that kernel auditing is broken when used with systemd's | |
126 | container code. When using systemd in conjunction with | |
127 | containers, please make sure to either turn off auditing at | |
128 | runtime using the kernel command line option "audit=0", or | |
129 | turn it off at kernel compile time using: | |
130 | CONFIG_AUDIT=n | |
131 | If systemd is compiled with libseccomp support on | |
132 | architectures which do not use socketcall() and where seccomp | |
133 | is supported (this effectively means x86-64 and ARM, but | |
134 | excludes 32-bit x86!), then nspawn will now install a | |
135 | work-around seccomp filter that makes containers boot even | |
136 | with audit being enabled. This works correctly only on kernels | |
137 | 3.14 and newer though. TL;DR: turn audit off, still. | |
138 | ||
139 | glibc >= 2.16 | |
140 | libcap | |
141 | libmount >= 2.30 (from util-linux) | |
142 | (util-linux *must* be built without --enable-libmount-support-mtab) | |
143 | libseccomp >= 2.3.1 (optional) | |
144 | libblkid >= 2.24 (from util-linux) (optional) | |
145 | libkmod >= 15 (optional) | |
146 | PAM >= 1.1.2 (optional) | |
147 | libcryptsetup (optional) | |
148 | libaudit (optional) | |
149 | libacl (optional) | |
150 | libselinux (optional) | |
151 | liblzma (optional) | |
152 | liblz4 >= 1.3.0 / 130 (optional) | |
153 | libzstd >= 1.4.0 (optional) | |
154 | libgcrypt (optional) | |
155 | libqrencode (optional) | |
156 | libmicrohttpd (optional) | |
157 | libpython (optional) | |
158 | libidn2 or libidn (optional) | |
159 | gnutls >= 3.1.4 (optional, >= 3.6.0 is required to support DNS-over-TLS with gnutls) | |
160 | openssl >= 1.1.0 (optional, required to support DNS-over-TLS with openssl) | |
161 | elfutils >= 158 (optional) | |
162 | polkit (optional) | |
163 | tzdata >= 2014f (optional) | |
164 | pkg-config | |
165 | gperf | |
166 | docbook-xsl (optional, required for documentation) | |
167 | xsltproc (optional, required for documentation) | |
168 | python-lxml (optional, required to build the indices) | |
169 | python >= 3.5 | |
170 | meson >= 0.46 (>= 0.49 is required to build position-independent executables) | |
171 | ninja | |
172 | gcc, awk, sed, grep, m4, and similar tools | |
173 | ||
174 | During runtime, you need the following additional | |
175 | dependencies: | |
176 | ||
177 | util-linux >= v2.27.1 required | |
178 | dbus >= 1.4.0 (strictly speaking optional, but recommended) | |
179 | NOTE: If using dbus < 1.9.18, you should override the default | |
180 | policy directory (--with-dbuspolicydir=/etc/dbus-1/system.d). | |
181 | dracut (optional) | |
182 | polkit (optional) | |
183 | ||
184 | To build in directory build/: | |
185 | meson build/ && ninja -C build | |
186 | ||
187 | Any configuration options can be specified as -Darg=value... arguments | |
188 | to meson. After the build directory is initially configured, meson will | |
189 | refuse to run again, and options must be changed with: | |
190 | mesonconf -Darg=value... | |
191 | mesonconf without any arguments will print out available options and | |
192 | their current values. | |
193 | ||
194 | Useful commands: | |
195 | ninja -v some/target | |
196 | ninja test | |
197 | sudo ninja install | |
198 | DESTDIR=... ninja install | |
199 | ||
200 | A tarball can be created with: | |
201 | git archive --format=tar --prefix=systemd-222/ v222 | xz > systemd-222.tar.xz | |
202 | ||
203 | When systemd-hostnamed is used, it is strongly recommended to | |
204 | install nss-myhostname to ensure that, in a world of | |
205 | dynamically changing hostnames, the hostname stays resolvable | |
206 | under all circumstances. In fact, systemd-hostnamed will warn | |
207 | if nss-myhostname is not installed. | |
208 | ||
209 | nss-systemd must be enabled on systemd systems, as that's required for | |
210 | DynamicUser= to work. Note that we ship services out-of-the-box that | |
211 | make use of DynamicUser= now, hence enabling nss-systemd is not | |
212 | optional. | |
213 | ||
214 | Note that the build prefix for systemd must be /usr. (Moreover, | |
215 | packages systemd relies on — such as D-Bus — really should use the same | |
216 | prefix, otherwise you are on your own.) -Dsplit-usr=false (which is the | |
217 | default and does not need to be specified) is the recommended setting, | |
218 | and -Dsplit-usr=true should be used on systems which have /usr on a | |
219 | separate partition. | |
220 | ||
221 | Additional packages are necessary to run some tests: | |
222 | - busybox (used by test/TEST-13-NSPAWN-SMOKE) | |
223 | - nc (used by test/TEST-12-ISSUE-3171) | |
224 | - python3-pyparsing | |
225 | - python3-evdev (used by hwdb parsing tests) | |
226 | - strace (used by test/test-functions) | |
227 | - capsh (optional, used by test-execute) | |
228 | ||
229 | USERS AND GROUPS: | |
230 | Default udev rules use the following standard system group | |
231 | names, which need to be resolvable by getgrnam() at any time, | |
232 | even in the very early boot stages, where no other databases | |
233 | and network are available: | |
234 | ||
235 | audio, cdrom, dialout, disk, input, kmem, kvm, lp, render, tape, tty, video | |
236 | ||
237 | During runtime, the journal daemon requires the | |
238 | "systemd-journal" system group to exist. New journal files will | |
239 | be readable by this group (but not writable), which may be used | |
240 | to grant specific users read access. In addition, system | |
241 | groups "wheel" and "adm" will be given read-only access to | |
242 | journal files using systemd-tmpfiles.service. | |
243 | ||
244 | The journal remote daemon requires the | |
245 | "systemd-journal-remote" system user and group to | |
246 | exist. During execution this network facing service will drop | |
247 | privileges and assume this uid/gid for security reasons. | |
248 | ||
249 | Similarly, the network management daemon requires the | |
250 | "systemd-network" system user and group to exist. | |
251 | ||
252 | Similarly, the name resolution daemon requires the | |
253 | "systemd-resolve" system user and group to exist. | |
254 | ||
255 | Similarly, the coredump support requires the | |
256 | "systemd-coredump" system user and group to exist. | |
257 | ||
258 | NSS: | |
259 | systemd ships with four glibc NSS modules: | |
260 | ||
261 | nss-myhostname resolves the local hostname to locally | |
262 | configured IP addresses, as well as "localhost" to | |
263 | 127.0.0.1/::1. | |
264 | ||
265 | nss-resolve enables DNS resolution via the systemd-resolved | |
266 | DNS/LLMNR caching stub resolver "systemd-resolved". | |
267 | ||
268 | nss-mymachines enables resolution of all local containers registered | |
269 | with machined to their respective IP addresses. It also maps UID/GIDs | |
270 | ranges used by containers to useful names. | |
271 | ||
272 | nss-systemd enables resolution of all dynamically allocated service | |
273 | users. (See the DynamicUser= setting in unit files.) | |
274 | ||
275 | To make use of these NSS modules, please add them to the "hosts:", | |
276 | "passwd:" and "group:" lines in /etc/nsswitch.conf. The "resolve" | |
277 | module should replace the glibc "dns" module in this file (and don't | |
278 | worry, it chain-loads the "dns" module if it can't talk to resolved). | |
279 | ||
280 | The four modules should be used in the following order: | |
281 | ||
282 | passwd: compat mymachines systemd | |
283 | group: compat mymachines systemd | |
284 | hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname | |
285 | ||
286 | SYSV INIT.D SCRIPTS: | |
287 | When calling "systemctl enable/disable/is-enabled" on a unit which is a | |
288 | SysV init.d script, it calls /usr/lib/systemd/systemd-sysv-install; | |
289 | this needs to translate the action into the distribution specific | |
290 | mechanism such as chkconfig or update-rc.d. Packagers need to provide | |
291 | this script if you need this functionality (you don't if you disabled | |
292 | SysV init support). | |
293 | ||
294 | Please see src/systemctl/systemd-sysv-install.SKELETON for how this | |
295 | needs to look like, and provide an implementation at the marked places. | |
296 | ||
297 | WARNINGS: | |
298 | systemd will warn during early boot if /usr is not already mounted at | |
299 | this point (that means: either located on the same file system as / or | |
300 | already mounted in the initrd). While in systemd itself very little | |
301 | will break if /usr is on a separate, late-mounted partition, many of | |
302 | its dependencies very likely will break sooner or later in one form or | |
303 | another. For example, udev rules tend to refer to binaries in /usr, | |
304 | binaries that link to libraries in /usr or binaries that refer to data | |
305 | files in /usr. Since these breakages are not always directly visible, | |
306 | systemd will warn about this, since this kind of file system setup is | |
307 | not really supported anymore by the basic set of Linux OS components. | |
308 | ||
309 | systemd requires that the /run mount point exists. systemd also | |
310 | requires that /var/run is a symlink to /run. | |
311 | ||
312 | For more information on this issue consult | |
313 | https://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken | |
314 | ||
315 | To run systemd under valgrind, compile with meson option | |
316 | -Dvalgrind=true and have valgrind development headers installed | |
317 | (i.e. valgrind-devel or equivalent). Otherwise, false positives will be | |
318 | triggered by code which violates some rules but is actually safe. Note | |
319 | that valgrind generates nice output only on exit(), hence on shutdown | |
320 | we don't execve() systemd-shutdown. | |
321 | ||
322 | STABLE BRANCHES AND BACKPORTS: | |
323 | Stable branches with backported patches are available in the | |
324 | systemd-stable repo at https://github.com/systemd/systemd-stable. | |
325 | ||
326 | Stable branches are started for certain releases of systemd and named | |
327 | after them, e.g. v238-stable. Stable branches are managed by | |
328 | distribution maintainers on an as needed basis. See | |
329 | https://www.freedesktop.org/wiki/Software/systemd/Backports/ for some | |
330 | more information and examples. | |
331 | ||
332 | ENGINEERING AND CONSULTING SERVICES: | |
333 | Kinvolk (https://kinvolk.io) offers professional engineering | |
334 | and consulting services for systemd. Please contact Chris Kühl | |
335 | <chris@kinvolk.io> for more information. |