`STOPPING=1` is final, explain in the sd_notify man page that this is
the case and log an error if a service attempts to get back out of
STOPPING with another notification.
Fixes#43465 / 98734ac74d
We nowadays expose pidfdid at various places, e.g. envvars
and dbus properties. Also the sd_notify() MAINPID= message
has been complemented with MAINPIDFDID=. But acquiring
pidfdid is actually non-trivial especially considering
the 32-bit case, hence let's introduce a public helper
in sd-daemon specifically for that purpose.
We have RestartMaxDelaySec= + RestartSteps= to exponentially increase
auto restart durations, but it currently cannot be reset by the service
itself, which makes it sometimes awkward to use. A typical pattern
in real life is that a service was once down (e.g. due to temporary
network interruption) and multiple restarts were attempted. Then,
future restarts would always wait for increated amount of time based on
RestartMaxDelaySec=, even after the original problem got resolved.
Such "persistence" could result in longer unavailablity than there
should be for failures that come later.
(C.f. https://utcc.utoronto.ca/~cks/space/blog/linux/SystemdResettingUnitBackoff)
Let's introduce a new sd_notify() notification for resetting the restart
counter. There were discussions about making this timer-based, but I think
it's more flexible to leave the decision-making to the service. This enables
them to do a combination of N successful requests + uptime check for instance.
Let's clarify more explicitly that privileged calls to
systemd-notify --pid= and sd_pid_notify() effectively override any
configured NotifyAccess=main|exec for a service.
In the troff output, this doesn't seem to make any difference. But in the
html output, the whitespace is sometimes preserved, creating an additional
gap before the following content. Drop it everywhere to avoid this.
The previous commit removed the UINT_MAX check for the fd array. Let's
now re-add one, but at a better place, and with a more useful limit. As
it turns out the kernel does not allow passing more than 253 fds at the
same time, hence use that as limit. And do so immediately before
calculating the control buffer size, so that we catch multiplication
overflows.
We are saying in public that the protocl is stable and can be easily
reimplemented, so provide an example doing so in the documentation,
license as MIT-0 so that it can be copied and pasted at will.
"submitted" is already used in the description of FDNAME=.
Let's use that instead of "stored" for FDPOLL= too, to make
it more clear that it's a per-submission/per-fdset setting.
In principle, arbitrary notifications may be sent via sd_notify. But in
practice, this is not useful at all, since the manager only accepts
notifications from services and ignores anything except a few specific
ones. The others will be logged if debugging is enabled. OTOH, the manager
produces EXIT_STATUS, but nothing in systemd looks at it, which is rather
confusing.
So remove the recommendation to use X_ prefixes, and instead say that other
messages will be ignored. Also, mention that mkosi uses this. Having an example
may be useful to understand what is going on.
Strangely, this is the first reference to mkosi in our man pages. Even more
strangely, debian is the only place which hosts the mkosi man page (among
the sites we have definitions for), so I linked to that version.
This tries to add information about when each option was added. It goes
back to version 183.
The version info is included from a separate file to allow generating it,
which would allow more control on the formatting of the final output.
Doesn't really matter since the two unicode symbols are supposedly
equivalent, but let's better follow the unicode recommendations to
prefer greek small letter mu, as per:
https://www.unicode.org/reports/tr25
I guess it was only a question of time until we need to add the final
frontier of notification functions: one that combines the features of
all the others:
1. specifiying a source PID
2. taking a list of fds to send along
3. accepting a format string for the status string
Hence, let's add it.
See #26688: getenv() is not thread-safe, and could a possible source of
problems when a multi-threaded program calls setenv()/putenv()/unsetenv() in
parallel. It is not possible to avoid getenv() calls in general, since $PATH,
$LANG, $SHELL, $USER, $HOME, $TZ may need to be accessed at any time.
Add a warning to our docs so that people are aware of the issue.
Closes#26688. (Real fixes will need to be in glibc and gnome-shell or other
programs.)
The text is added to threads-aware.xml to be included in various places. By
including it in libsystemd-pkgconfig.xml, it is automatically added to all sd-*
pages. The text is also included explicitly in pages for a few other functions
which are call getenv().
Whether we use a newline at the beginning of <programlisting> or not doesn't
seem to have any effect. So let's use the newline in multi-line examples; it is
easier to get the indentation right this way.
Allow sending notifications via AF_VSOCK, so that VMs can communicate
to the hypervisor/VMM that they are finished booting.
Note that if the hypervisor does not support SOCK_DGRAM over AF_VSOCK
(ie: qemu at the time of writing), SOCK_SEQPACKET will be used instead.
Let's nudge people towards the use of an anonymous buffer like we
do internally.
"errno" → "errnum", to match the man page for strerror, and also to avoid
confusion with the global variable. In general, I think that errno is a
terrible interface and we shouldn't encourage people to use it. Those functions
use errno-style error numbers, which are a different thing.
This adds the sd_notify_barrier function, to allow users to synchronize against
the reception of sd_notify(3) status messages. It acts as a synchronization
point, and a successful return gurantees that all previous messages have been
consumed by the manager. This can be used to eliminate race conditions where
the sending process exits too early for systemd to associate its PID to a
cgroup and attribute the status message to a unit correctly.
systemd-notify now uses this function for proper notification delivery and be
useful for NotifyAccess=all units again in user mode, or in cases where it
doesn't have a control process as parent.
Fixes: #2739
A service can specify FDSTORE=1 FDPOLL=0 to request that PID1 does not
poll the fd to remove them on error. If set, fds will only be removed on
FDSTOREREMOVE=1 or when the service is done.
Fixes: #12086
David 'equinox' Lamparter
Luca Boccassi
Mike Yuan
Daan De Meyer
Jan Engelhardt
Zbigniew Jędrzejewski-Szmek
Lennart Poettering
Dmitry V. Levin
Luca Boccassi
Simon Fowler
David Tardon
Abderrahim Kitouni
Yu Watanabe
Frantisek Sumsal
Kumar Kartikeya Dwivedi
Kenny Levinsen
Philip Withnall