On the unified hierarchy, blkio controller is renamed to io and the interface
is changed significantly.
* blkio.weight and blkio.weight_device are consolidated into io.weight which
uses the standardized weight range [1, 10000] with 100 as the default value.
* blkio.throttle.{read|write}_{bps|iops}_device are consolidated into io.max.
Expansion of throttling features is being worked on to support
work-conserving absolute limits (io.low and io.high).
* All stats are consolidated into io.stats.
This patchset adds support for the new interface. As the interface has been
revamped and new features are expected to be added, it seems best to treat it
as a separate controller rather than trying to expand the blkio settings
although we might add automatic translation if only blkio settings are
specified.
* io.weight handling is mostly identical to blkio.weight[_device] handling
except that the weight range is different.
* Both read and write bandwidth settings are consolidated into
CGroupIODeviceLimit which describes all limits applicable to the device.
This makes it less painful to add new limits.
* "max" can be used to specify the maximum limit which is equivalent to no
config for max limits and treated as such. If a given CGroupIODeviceLimit
doesn't contain any non-default configs, the config struct is discarded once
the no limit config is applied to cgroup.
* lookup_blkio_device() is renamed to lookup_block_device().
Signed-off-by: Tejun Heo <htejun@fb.com>
dbus-daemon currently uses a backlog of 30 on its D-bus system bus socket. On
overloaded systems this means that only 30 connections may be queued without
dbus-daemon processing them before further connection attempts fail. Our
cgroups-agent binary so far used D-Bus for its messaging, and hitting this
limit hence may result in us losing cgroup empty messages.
This patch adds a seperate cgroup agent socket of type AF_UNIX/SOCK_DGRAM.
Since sockets of these types need no connection set up, no listen() backlog
applies. Our cgroup-agent binary will hence simply block as long as it can't
enqueue its datagram message, so that we won't lose cgroup empty messages as
likely anymore.
This also rearranges the ordering of the processing of SIGCHLD signals, service
notification messages (sd_notify()...) and the two types of cgroup
notifications (inotify for the unified hierarchy support, and agent for the
classic hierarchy support). We now always process events for these in the
following order:
1. service notification messages (SD_EVENT_PRIORITY_NORMAL-7)
2. SIGCHLD signals (SD_EVENT_PRIORITY_NORMAL-6)
3. cgroup inotify and cgroup agent (SD_EVENT_PRIORITY_NORMAL-5)
This is because when receiving SIGCHLD we invalidate PID information, which we
need to process the service notification messages which are bound to PIDs.
Hence the order between the first two items. And we want to process SIGCHLD
metadata to detect whether a service is gone, before using cgroup
notifications, to decide when a service is gone, since the former carries more
useful metadata.
Related to this:
https://bugs.freedesktop.org/show_bug.cgi?id=95264https://github.com/systemd/systemd/issues/1961
unit_has_mask_realized() determines whether the specified unit has its cgroups
set up properly given the desired target_mask; however, on the unified
hierarchy, controllers need to be enabled explicitly for children and the mask
of enabled controllers can deviate from target_mask. Only considering
target_mask in unit_has_mask_realized() can lead to false positives and
skipping enabling the requested controllers.
This patch adds unit->cgroup_enabled_mask to track which controllers are
enabled and updates unit_has_mask_realized() to also consider enable_mask.
Signed-off-by: Tejun Heo <htejun@fb.com>
Previously, we had two enums ManagerRunningAs and UnitFileScope, that were
mostly identical and converted from one to the other all the time. The latter
had one more value UNIT_FILE_GLOBAL however.
Let's simplify things, and remove ManagerRunningAs and replace it by
UnitFileScope everywhere, thus making the translation unnecessary. Introduce
two new macros MANAGER_IS_SYSTEM() and MANAGER_IS_USER() to simplify checking
if we are running in one or the user context.
Earlier during the development of unified hierarchy, the populated event was
reported through by the dedicated "cgroup.populated" file; however, the
interface was updated so that it's reported through the "populated" field of
"cgroup.events" file. Update populated event handling logic accordingly.
Support for net_cls.class_id through the NetClass= configuration directive
has been added in v227 in preparation for a per-unit packet filter mechanism.
However, it turns out the kernel people have decided to deprecate the net_cls
and net_prio controllers in v2. Tejun provides a comprehensive justification
for this in his commit, which has landed during the merge window for kernel
v4.5:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
As we're aiming for full support for the v2 cgroup hierarchy, we can no
longer support this feature. Userspace tool such as nftables are moving over
to setting rules that are specific to the full cgroup path of a task, which
obsoletes these controllers anyway.
This commit removes support for tweaking details in the net_cls controller,
but keeps the NetClass= directive around for legacy compatibility reasons.
There are more than enough calls doing string manipulations to deserve
its own files, hence do something about it.
This patch also sorts the #include blocks of all files that needed to be
updated, according to the sorting suggestions from CODING_STYLE. Since
pretty much every file needs our string manipulation functions this
effectively means that most files have sorted #include blocks now.
Also touches a few unrelated include files.
Add a new config directive called NetClass= to CGroup enabled units.
Allowed values are positive numbers for fix assignments and "auto" for
picking a free value automatically, for which we need to keep track of
dynamically assigned net class IDs of units. Introduce a hash table for
this, and also record the last ID that was given out, so the allocator
can start its search for the next 'hole' from there. This could
eventually be optimized with something like an irb.
The class IDs up to 65536 are considered reserved and won't be
assigned automatically by systemd. This barrier can be made a config
directive in the future.
Values set in unit files are stored in the CGroupContext of the
unit and considered read-only. The actually assigned number (which
may have been chosen dynamically) is stored in the unit itself and
is guaranteed to remain stable as long as the unit is active.
In the CGroup controller, set the configured CGroup net class to
net_cls.classid. Multiple unit may share the same net class ID,
and those which do are linked together.
Let's make sure that we follow the same codepaths when adjusting a
cgroup property via the dbus SetProperty() call, and when we execute the
StartupCPUShares= effect.
Let's stop using the "unsigned long" type for weights/shares, and let's
just use uint64_t for this, as that's what we expose on the bus.
Unify parsers, and always validate the range for these fields.
Correct the default blockio weight to 500, since that's what the kernel
actually uses.
When parsing the weight/shares settings from unit files accept the empty
string as a way to reset the weight/shares value. When getting it via
the bus, uniformly map (uint64_t) -1 to unset.
Open up StartupCPUShares= and StartupBlockIOWeight= to transient units.
This adds support for the new "pids" cgroup controller of 4.3 kernels.
It allows accounting the number of tasks in a cgroup and enforcing
limits on it.
This adds two new setting TasksAccounting= and TasksMax= to each unit,
as well as a gloabl option DefaultTasksAccounting=.
This also updated "cgtop" to optionally make use of the new
kernel-provided accounting.
systemctl has been updated to show the number of tasks for each service
if it is available.
This patch also adds correct support for undoing memory limits for units
using a MemoryLimit=infinity syntax. We do the same for TasksMax= now
and hence keep things in sync here.
Delegation to unpriviliged processes is safe in the unified hierarchy,
hence allow it. This has the benefit of permitting "systemd --user"
instances to further partition their resources between user services.
Let's move the actual cgroup part of it into a new separate function
manager_get_unit_by_pid_cgroup(), and then make
manager_get_unit_by_pid() just a wrapper that also checks the two pid
hashmaps.
Then, let's make sure the various calls that want to deliver events to
the owners of a PID check both hashmaps and the cgroup and deliver the
event to *each* of them. OTOH make sure bus calls like GetUnitByPID()
continue to check the PID hashmaps first and the cgroup only as
fallback.
This adds a new PID_TO_PTR() macro, plus PTR_TO_PID() and makes use of
it wherever we maintain processes in a hash table. Previously we
sometimes used LONG_TO_PTR() and other times ULONG_TO_PTR() for that,
hence let's make this more explicit and clean up things.
This patch set adds full support the new unified cgroup hierarchy logic
of modern kernels.
A new kernel command line option "systemd.unified_cgroup_hierarchy=1" is
added. If specified the unified hierarchy is mounted to /sys/fs/cgroup
instead of a tmpfs. No further hierarchies are mounted. The kernel
command line option defaults to off. We can turn it on by default as
soon as the kernel's APIs regarding this are stabilized (but even then
downstream distros might want to turn this off, as this will break any
tools that access cgroupfs directly).
It is possibly to choose for each boot individually whether the unified
or the legacy hierarchy is used. nspawn will by default provide the
legacy hierarchy to containers if the host is using it, and the unified
otherwise. However it is possible to run containers with the unified
hierarchy on a legacy host and vice versa, by setting the
$UNIFIED_CGROUP_HIERARCHY environment variable for nspawn to 1 or 0,
respectively.
The unified hierarchy provides reliable cgroup empty notifications for
the first time, via inotify. To make use of this we maintain one
manager-wide inotify fd, and each cgroup to it.
This patch also removes cg_delete() which is unused now.
On kernel 4.2 only the "memory" controller is compatible with the
unified hierarchy, hence that's the only controller systemd exposes when
booted in unified heirarchy mode.
This introduces a new enum for enumerating supported controllers, plus a
related enum for the mask bits mapping to it. The core is changed to
make use of this everywhere.
This moves PID 1 into a new "init.scope" implicit scope unit in the root
slice. This is necessary since on the unified hierarchy cgroups may
either contain subgroups or processes but not both. PID 1 hence has to
move out of the root cgroup (strictly speaking the root cgroup is the
only one where processes and subgroups are still allowed, but in order
to support containers nicey, we move PID 1 into the new scope in all
cases.) This new unit is also used on legacy hierarchy setups. It's
actually pretty useful on all systems, as it can then be used to filter
journal messages coming from PID 1, and so on.
The root slice ("-.slice") is now implicitly created and started (and
does not require a unit file on disk anymore), since
that's where "init.scope" is located and the slice needs to be started
before the scope can.
To check whether we are in unified or legacy hierarchy mode we use
statfs() on /sys/fs/cgroup. If the .f_type field reports tmpfs we are in
legacy mode, if it reports cgroupfs we are in unified mode.
This patch set carefuly makes sure that cgls and cgtop continue to work
as desired.
When invoking nspawn as a service it will implicitly create two
subcgroups in the cgroup it is using, one to move the nspawn process
into, the other to move the actual container processes into. This is
done because of the requirement that cgroups may either contain
processes or other subgroups.
It's cheaper that going to cgroupfs, and also usually the better choice
since it's not racy and can map PIDs even if they were moved to a
different unit.
In all cases where the function (or cg_is_empty_recursive()) ignoring
the calling process is actually wrong, as a process keeps a cgroup busy
regardless if its the current one or another. Hence, let's simplify
things and drop the "ignore_self" parameter.
The legacy cgroup hierarchy does not support reliable empty
notifications in containers and if there are left-over subgroups in a
cgroup. This makes it hard to correctly wait for them running empty, and
thus we previously disabled this logic entirely.
With this change we explicitly check for the container case, and whether
the unit is a "delegation" unit (i.e. one where programs may create
their own subgroups). If we are neither in a container, nor operating on
a delegation unit cgroup empty notifications become reliable and thus we
start waiting for the empty notifications again.
This doesn't really fix the general problem around cgroup notifications
but reduces the effect around it.
(This also reorders #include lines by their focus, as suggsted in
CODING_STYLE. We have to add "virt.h", so let's do that at the right
place.)
Also see #317.
It's primarily just a property of the Manager object after all, and we
try to refer to PID 1 as "manager" instead of "systemd", hence let's to
stick to this here too.
If a cgroup fails to be destroyed (most likely because there are still
processes running as part of a service after the main pid exits), don't
free and remove the cgroup unit from the manager. This fixes a
regression introduced by the cgroup rework in v205 where systemd would
forget about processes still running after the unit becomes inactive.
(This can happen when the main pid exits and KillMode=process or none).
If the format string contains %m, clearly errno must have a meaningful
value, so we might as well use log_*_errno to have ERRNO= logged.
Using:
find . -name '*.[ch]' | xargs sed -r -i -e \
's/log_(debug|info|notice|warning|error|emergency)\((".*%m.*")/log_\1_errno(errno, \2/'
Plus some whitespace, linewrap, and indent adjustments.
As a followup to 086891e5c1 "log: add an "error" parameter to all
low-level logging calls and intrdouce log_error_errno() as log calls
that take error numbers", use sed to convert the simple cases to use
the new macros:
find . -name '*.[ch]' | xargs sed -r -i -e \
's/log_(debug|info|notice|warning|error|emergency)\("(.*)%s"(.*), strerror\(-([a-zA-Z_]+)\)\);/log_\1_errno(-\4, "\2%m"\3);/'
Multi-line log_*() invocations are not covered.
And we also should add log_unit_*_errno().
For priviliged units this resource control property ensures that the
processes have all controllers systemd manages enabled.
For unpriviliged services (those with User= set) this ensures that
access rights to the service cgroup is granted to the user in question,
to create further subgroups. Note that this only applies to the
name=systemd hierarchy though, as access to other controllers is not
safe for unpriviliged processes.
Delegate=yes should be set for container scopes where a systemd instance
inside the container shall manage the hierarchies below its own cgroup
and have access to all controllers.
Delegate=yes should also be set for user@.service, so that systemd
--user can run, controlling its own cgroup tree.
This commit changes machined, systemd-nspawn@.service and user@.service
to set this boolean, in order to ensure that container management will
just work, and the user systemd instance can run fine.
systemctl would print 'CPUQuotaPerSecUSec=(null)' for no limit. This
does not look right.
Since USEC_INFINITY is one of the valid values, format_timespan()
could return NULL, and we should wrap every use of it in strna() or
similar. But most callers didn't do that, and it seems more robust to
return a string ("infinity") that makes sense most of the time, even
if in some places the result will not be grammatically correct.
We'll stay in "initializing" until basic.target has reached, at which
point we will enter "starting".
This is preparation so that we can change the startip timeout to only
apply to the first phase of startup, not the full procedure.
Also add a bit of debugging output to help diagnose problems,
add missing units, and simplify cppflags.
Move test-engine to normal tests from manual tests, it should now
work without destroying the system.
Only accept cpu quota values in percentages, get rid of period
definition.
It's not clear whether the CFS period controllable per-cgroup even has a
future in the kernel, hence let's simplify all this, hardcode the period
to 100ms and only accept percentage based quota values.
Similar to CPUShares= and BlockIOWeight= respectively. However only
assign the specified weight during startup. Each control group
attribute is re-assigned as weight by CPUShares=weight and
BlockIOWeight=weight after startup. If not CPUShares= or
BlockIOWeight= be specified, then the attribute is re-assigned to each
default attribute value. (default cpu.shares=1024, blkio.weight=1000)
If only CPUShares=weight or BlockIOWeight=weight be specified, then
that implies StartupCPUShares=weight and StartupBlockIOWeight=weight.
safe_close() automatically becomes a NOP when a negative fd is passed,
and returns -1 unconditionally. This makes it easy to write lines like
this:
fd = safe_close(fd);
Which will close an fd if it is open, and reset the fd variable
correctly.
By making use of this new scheme we can drop a > 200 lines of code that
was required to test for non-negative fds or to reset the closed fd
variable afterwards.
Previously a cgroup setting down tree would result in cgroup membership
additions being propagated up the tree and to the siblings, however a
unit could never lose cgroup memberships again. With this change we'll
make sure that both cgroup additions and removals propagate properly.
This way cleaning up the cgroup tree on shutdown is a lot easier since
we are in the root dir. Also PID 1 was previously artificially placed in
system.slice, even though our rule actually was not to have processes in
slices. The root slice otoh is magic anyway, so having PID 1 in there
sounds less surprising.
Of course, this means that PID is scheduled against the three top-level
slices.
Previously we did operations like attach, trim or migrate only on the
controllers that were enabled for a specific unit. With this changes we
will now do them for all supproted controllers, and fall back to all
possible prefix paths if the specified paths do not exist.
This fixes issues if a controller is being disabled for a unit where it
was previously enabled, and makes sure that all processes stay as "far
down" the tree as groups exist.
The non-hierarchial mode contradicts the whole idea of a cgroup tree so
let's not support this. In the future the kernel will only support the
hierarchial logic anyway.
The cgroup attribute memory.soft_limit_in_bytes is unlikely to stay
around in the kernel for good, so let's not expose it for now. We can
readd something like it later when the kernel guys decided on a final
API for this.
If the memory_limit of unit is -1, we should write "-1"
to the file memory.limit_in_bytes. not the (unit64_t) -1.
otherwise the memory.limit_in_bytes will be set to zero.
Some units set KillMode=none to survive the initrd→rootfs transition. We
cannot remove their cgroups, but that shouldn't really be considered an
issue, so let's downgrade the error message.
Replace the very generic cgroup hookup with a much simpler one. With
this change only the high-level cgroup settings remain, the ability to
set arbitrary cgroup attributes is removed, so is support for adding
units to arbitrary cgroup controllers or setting arbitrary paths for
them (especially paths that are different for the various controllers).
This also introduces a new -.slice root slice, that is the parent of
system.slice and friends. This enables easy admin configuration of
root-level cgrouo properties.
This replaces DeviceDeny= by DevicePolicy=, and implicitly adds in
/dev/null, /dev/zero and friends if DeviceAllow= is used (unless this is
turned off by DevicePolicy=).
- This changes all logind cgroup objects to use slice objects rather
than fixed croup locations.
- logind can now collect minimal information about running
VMs/containers. As fixed cgroup locations can no longer be used we
need an entity that keeps track of machine cgroups in whatever slice
they might be located. Since logind already keeps track of users,
sessions and seats this is a trivial addition.
- nspawn will now register with logind and pass various bits of metadata
along. A new option "--slice=" has been added to place the container
in a specific slice.
- loginctl gained commands to list, introspect and terminate machines.
- user.slice and machine.slice will now be pulled in by logind.service,
since only logind.service requires this slice.
In order to prepare for the kernel cgroup rework, let's introduce a new
unit type to systemd, the "slice". Slices can be arranged in a tree and
are useful to partition resources freely and hierarchally by the user.
Each service unit can now be assigned to one of these slices, and later
on login users and machines may too.
Slices translate pretty directly to the cgroup hierarchy, and the
various objects can be assigned to any of the slices in the tree.
Containers will now carry a label (normally derived from the root
directory name, but configurable by the user), and the container's root
cgroup is /machine/<label>. This label is called "machine name", and can
cover both containers and VMs (as soon as libvirt also makes use of
/machine/).
libsystemd-login can be used to query the machine name from a process.
This patch also includes numerous clean-ups for the cgroup code.
This allows clients to put inotify watches on these trees to watch for
state changes, without having to wait until these dirs are created.
This introduces the new top-level /machine cgroup dir as canonical
location where OS containers and VMs shall be located (as discussed with
the libvirt folks).
Note: I did s/MANAGER/SYSTEMD/ everywhere, even though it makes the
patch quite verbose. Nevertheless, keeping MANAGER prefix in some
places, and SYSTEMD prefix in others would just lead to confusion down
the road. Better to rip off the band-aid now.
also a number of minor fixups and bug fixes: spelling, oom errors
that didn't print errors, not properly forwarding error codes,
few more consistency issues, et cetera
glibc/glib both use "out of memory" consistantly so maybe we should
consider that instead of this.
Eliminates one string out of a number of binaries. Also fixes extra newline
in udev/scsi_id
The kernel will only notify us of cgroups running empty if no subcgroups
exist anymore. Hence make sure we don't leave our own control/ subcgroup
around longer than necessary.
https://bugzilla.redhat.com/show_bug.cgi?id=818381
Previously, we were brutally and onconditionally killing all processes
in a service's cgroup before starting the service anew, in order to
ensure that StartPre lines cannot be misused to spawn long-running
processes.
On logind-less systems this has the effect that restarting sshd
necessarily calls all active ssh sessions, which is usually not
desirable.
With this patch control processes for a service are placed in a
sub-cgroup called "control/". When starting a service anew we simply
kill this cgroup, but not the main cgroup, in order to avoid killing any
long-running non-control processes from previous runs.
https://bugzilla.redhat.com/show_bug.cgi?id=805942
We finally got the OK from all contributors with non-trivial commits to
relicense systemd from GPL2+ to LGPL2.1+.
Some udev bits continue to be GPL2+ for now, but we are looking into
relicensing them too, to allow free copy/paste of all code within
systemd.
The bits that used to be MIT continue to be MIT.
The big benefit of the relicensing is that closed source code may now
link against libsystemd-login.so and friends.