8 :Author: Tejun Heo <tj@kernel.org>
10 This is the authoritative documentation on the design, interface and
11 conventions of cgroup v2. It describes all userland-visible aspects
12 of cgroup including core and specific controller behaviors. All
13 future changes must be reflected in this document. Documentation for
14 v1 is available under :ref:`Documentation/admin-guide/cgroup-v1/index.rst <cgroup-v1>`.
23 2-2. Organizing Processes and Threads
26 2-3. [Un]populated Notification
27 2-4. Controlling Controllers
28 2-4-1. Enabling and Disabling
29 2-4-2. Top-down Constraint
30 2-4-3. No Internal Process Constraint
32 2-5-1. Model of Delegation
33 2-5-2. Delegation Containment
35 2-6-1. Organize Once and Control
36 2-6-2. Avoid Name Collisions
37 3. Resource Distribution Models
45 4-3. Core Interface Files
48 5-1-1. CPU Interface Files
50 5-2-1. Memory Interface Files
51 5-2-2. Usage Guidelines
52 5-2-3. Memory Ownership
54 5-3-1. IO Interface Files
57 5-3-3-1. How IO Latency Throttling Works
58 5-3-3-2. IO Latency Interface Files
61 5-4-1. PID Interface Files
63 5.5-1. Cpuset Interface Files
66 5-7-1. RDMA Interface Files
68 5.8-1. HugeTLB Interface Files
70 5.9-1 Miscellaneous cgroup Interface Files
71 5.9-2 Migration and Ownership
74 5-N. Non-normative information
75 5-N-1. CPU controller root cgroup process behaviour
76 5-N-2. IO controller root cgroup process behaviour
79 6-2. The Root and Views
80 6-3. Migration and setns(2)
81 6-4. Interaction with Other Namespaces
82 P. Information on Kernel Programming
83 P-1. Filesystem Support for Writeback
84 D. Deprecated v1 Core Features
85 R. Issues with v1 and Rationales for v2
86 R-1. Multiple Hierarchies
87 R-2. Thread Granularity
88 R-3. Competition Between Inner Nodes and Threads
89 R-4. Other Interface Issues
90 R-5. Controller Issues and Remedies
100 "cgroup" stands for "control group" and is never capitalized. The
101 singular form is used to designate the whole feature and also as a
102 qualifier as in "cgroup controllers". When explicitly referring to
103 multiple individual control groups, the plural form "cgroups" is used.
109 cgroup is a mechanism to organize processes hierarchically and
110 distribute system resources along the hierarchy in a controlled and
113 cgroup is largely composed of two parts - the core and controllers.
114 cgroup core is primarily responsible for hierarchically organizing
115 processes. A cgroup controller is usually responsible for
116 distributing a specific type of system resource along the hierarchy
117 although there are utility controllers which serve purposes other than
118 resource distribution.
120 cgroups form a tree structure and every process in the system belongs
121 to one and only one cgroup. All threads of a process belong to the
122 same cgroup. On creation, all processes are put in the cgroup that
123 the parent process belongs to at the time. A process can be migrated
124 to another cgroup. Migration of a process doesn't affect already
125 existing descendant processes.
127 Following certain structural constraints, controllers may be enabled or
128 disabled selectively on a cgroup. All controller behaviors are
129 hierarchical - if a controller is enabled on a cgroup, it affects all
130 processes which belong to the cgroups consisting the inclusive
131 sub-hierarchy of the cgroup. When a controller is enabled on a nested
132 cgroup, it always restricts the resource distribution further. The
133 restrictions set closer to the root in the hierarchy can not be
134 overridden from further away.
143 Unlike v1, cgroup v2 has only single hierarchy. The cgroup v2
144 hierarchy can be mounted with the following mount command::
146 # mount -t cgroup2 none $MOUNT_POINT
148 cgroup2 filesystem has the magic number 0x63677270 ("cgrp"). All
149 controllers which support v2 and are not bound to a v1 hierarchy are
150 automatically bound to the v2 hierarchy and show up at the root.
151 Controllers which are not in active use in the v2 hierarchy can be
152 bound to other hierarchies. This allows mixing v2 hierarchy with the
153 legacy v1 multiple hierarchies in a fully backward compatible way.
155 A controller can be moved across hierarchies only after the controller
156 is no longer referenced in its current hierarchy. Because per-cgroup
157 controller states are destroyed asynchronously and controllers may
158 have lingering references, a controller may not show up immediately on
159 the v2 hierarchy after the final umount of the previous hierarchy.
160 Similarly, a controller should be fully disabled to be moved out of
161 the unified hierarchy and it may take some time for the disabled
162 controller to become available for other hierarchies; furthermore, due
163 to inter-controller dependencies, other controllers may need to be
166 While useful for development and manual configurations, moving
167 controllers dynamically between the v2 and other hierarchies is
168 strongly discouraged for production use. It is recommended to decide
169 the hierarchies and controller associations before starting using the
170 controllers after system boot.
172 During transition to v2, system management software might still
173 automount the v1 cgroup filesystem and so hijack all controllers
174 during boot, before manual intervention is possible. To make testing
175 and experimenting easier, the kernel parameter cgroup_no_v1= allows
176 disabling controllers in v1 and make them always available in v2.
178 cgroup v2 currently supports the following mount options.
181 Consider cgroup namespaces as delegation boundaries. This
182 option is system wide and can only be set on mount or modified
183 through remount from the init namespace. The mount option is
184 ignored on non-init namespace mounts. Please refer to the
185 Delegation section for details.
188 Only populate memory.events with data for the current cgroup,
189 and not any subtrees. This is legacy behaviour, the default
190 behaviour without this option is to include subtree counts.
191 This option is system wide and can only be set on mount or
192 modified through remount from the init namespace. The mount
193 option is ignored on non-init namespace mounts.
196 Recursively apply memory.min and memory.low protection to
197 entire subtrees, without requiring explicit downward
198 propagation into leaf cgroups. This allows protecting entire
199 subtrees from one another, while retaining free competition
200 within those subtrees. This should have been the default
201 behavior but is a mount-option to avoid regressing setups
202 relying on the original semantics (e.g. specifying bogusly
203 high 'bypass' protection values at higher tree levels).
206 Organizing Processes and Threads
207 --------------------------------
212 Initially, only the root cgroup exists to which all processes belong.
213 A child cgroup can be created by creating a sub-directory::
217 A given cgroup may have multiple child cgroups forming a tree
218 structure. Each cgroup has a read-writable interface file
219 "cgroup.procs". When read, it lists the PIDs of all processes which
220 belong to the cgroup one-per-line. The PIDs are not ordered and the
221 same PID may show up more than once if the process got moved to
222 another cgroup and then back or the PID got recycled while reading.
224 A process can be migrated into a cgroup by writing its PID to the
225 target cgroup's "cgroup.procs" file. Only one process can be migrated
226 on a single write(2) call. If a process is composed of multiple
227 threads, writing the PID of any thread migrates all threads of the
230 When a process forks a child process, the new process is born into the
231 cgroup that the forking process belongs to at the time of the
232 operation. After exit, a process stays associated with the cgroup
233 that it belonged to at the time of exit until it's reaped; however, a
234 zombie process does not appear in "cgroup.procs" and thus can't be
235 moved to another cgroup.
237 A cgroup which doesn't have any children or live processes can be
238 destroyed by removing the directory. Note that a cgroup which doesn't
239 have any children and is associated only with zombie processes is
240 considered empty and can be removed::
244 "/proc/$PID/cgroup" lists a process's cgroup membership. If legacy
245 cgroup is in use in the system, this file may contain multiple lines,
246 one for each hierarchy. The entry for cgroup v2 is always in the
249 # cat /proc/842/cgroup
251 0::/test-cgroup/test-cgroup-nested
253 If the process becomes a zombie and the cgroup it was associated with
254 is removed subsequently, " (deleted)" is appended to the path::
256 # cat /proc/842/cgroup
258 0::/test-cgroup/test-cgroup-nested (deleted)
264 cgroup v2 supports thread granularity for a subset of controllers to
265 support use cases requiring hierarchical resource distribution across
266 the threads of a group of processes. By default, all threads of a
267 process belong to the same cgroup, which also serves as the resource
268 domain to host resource consumptions which are not specific to a
269 process or thread. The thread mode allows threads to be spread across
270 a subtree while still maintaining the common resource domain for them.
272 Controllers which support thread mode are called threaded controllers.
273 The ones which don't are called domain controllers.
275 Marking a cgroup threaded makes it join the resource domain of its
276 parent as a threaded cgroup. The parent may be another threaded
277 cgroup whose resource domain is further up in the hierarchy. The root
278 of a threaded subtree, that is, the nearest ancestor which is not
279 threaded, is called threaded domain or thread root interchangeably and
280 serves as the resource domain for the entire subtree.
282 Inside a threaded subtree, threads of a process can be put in
283 different cgroups and are not subject to the no internal process
284 constraint - threaded controllers can be enabled on non-leaf cgroups
285 whether they have threads in them or not.
287 As the threaded domain cgroup hosts all the domain resource
288 consumptions of the subtree, it is considered to have internal
289 resource consumptions whether there are processes in it or not and
290 can't have populated child cgroups which aren't threaded. Because the
291 root cgroup is not subject to no internal process constraint, it can
292 serve both as a threaded domain and a parent to domain cgroups.
294 The current operation mode or type of the cgroup is shown in the
295 "cgroup.type" file which indicates whether the cgroup is a normal
296 domain, a domain which is serving as the domain of a threaded subtree,
297 or a threaded cgroup.
299 On creation, a cgroup is always a domain cgroup and can be made
300 threaded by writing "threaded" to the "cgroup.type" file. The
301 operation is single direction::
303 # echo threaded > cgroup.type
305 Once threaded, the cgroup can't be made a domain again. To enable the
306 thread mode, the following conditions must be met.
308 - As the cgroup will join the parent's resource domain. The parent
309 must either be a valid (threaded) domain or a threaded cgroup.
311 - When the parent is an unthreaded domain, it must not have any domain
312 controllers enabled or populated domain children. The root is
313 exempt from this requirement.
315 Topology-wise, a cgroup can be in an invalid state. Please consider
316 the following topology::
318 A (threaded domain) - B (threaded) - C (domain, just created)
320 C is created as a domain but isn't connected to a parent which can
321 host child domains. C can't be used until it is turned into a
322 threaded cgroup. "cgroup.type" file will report "domain (invalid)" in
323 these cases. Operations which fail due to invalid topology use
324 EOPNOTSUPP as the errno.
326 A domain cgroup is turned into a threaded domain when one of its child
327 cgroup becomes threaded or threaded controllers are enabled in the
328 "cgroup.subtree_control" file while there are processes in the cgroup.
329 A threaded domain reverts to a normal domain when the conditions
332 When read, "cgroup.threads" contains the list of the thread IDs of all
333 threads in the cgroup. Except that the operations are per-thread
334 instead of per-process, "cgroup.threads" has the same format and
335 behaves the same way as "cgroup.procs". While "cgroup.threads" can be
336 written to in any cgroup, as it can only move threads inside the same
337 threaded domain, its operations are confined inside each threaded
340 The threaded domain cgroup serves as the resource domain for the whole
341 subtree, and, while the threads can be scattered across the subtree,
342 all the processes are considered to be in the threaded domain cgroup.
343 "cgroup.procs" in a threaded domain cgroup contains the PIDs of all
344 processes in the subtree and is not readable in the subtree proper.
345 However, "cgroup.procs" can be written to from anywhere in the subtree
346 to migrate all threads of the matching process to the cgroup.
348 Only threaded controllers can be enabled in a threaded subtree. When
349 a threaded controller is enabled inside a threaded subtree, it only
350 accounts for and controls resource consumptions associated with the
351 threads in the cgroup and its descendants. All consumptions which
352 aren't tied to a specific thread belong to the threaded domain cgroup.
354 Because a threaded subtree is exempt from no internal process
355 constraint, a threaded controller must be able to handle competition
356 between threads in a non-leaf cgroup and its child cgroups. Each
357 threaded controller defines how such competitions are handled.
360 [Un]populated Notification
361 --------------------------
363 Each non-root cgroup has a "cgroup.events" file which contains
364 "populated" field indicating whether the cgroup's sub-hierarchy has
365 live processes in it. Its value is 0 if there is no live process in
366 the cgroup and its descendants; otherwise, 1. poll and [id]notify
367 events are triggered when the value changes. This can be used, for
368 example, to start a clean-up operation after all processes of a given
369 sub-hierarchy have exited. The populated state updates and
370 notifications are recursive. Consider the following sub-hierarchy
371 where the numbers in the parentheses represent the numbers of processes
377 A, B and C's "populated" fields would be 1 while D's 0. After the one
378 process in C exits, B and C's "populated" fields would flip to "0" and
379 file modified events will be generated on the "cgroup.events" files of
383 Controlling Controllers
384 -----------------------
386 Enabling and Disabling
387 ~~~~~~~~~~~~~~~~~~~~~~
389 Each cgroup has a "cgroup.controllers" file which lists all
390 controllers available for the cgroup to enable::
392 # cat cgroup.controllers
395 No controller is enabled by default. Controllers can be enabled and
396 disabled by writing to the "cgroup.subtree_control" file::
398 # echo "+cpu +memory -io" > cgroup.subtree_control
400 Only controllers which are listed in "cgroup.controllers" can be
401 enabled. When multiple operations are specified as above, either they
402 all succeed or fail. If multiple operations on the same controller
403 are specified, the last one is effective.
405 Enabling a controller in a cgroup indicates that the distribution of
406 the target resource across its immediate children will be controlled.
407 Consider the following sub-hierarchy. The enabled controllers are
408 listed in parentheses::
410 A(cpu,memory) - B(memory) - C()
413 As A has "cpu" and "memory" enabled, A will control the distribution
414 of CPU cycles and memory to its children, in this case, B. As B has
415 "memory" enabled but not "CPU", C and D will compete freely on CPU
416 cycles but their division of memory available to B will be controlled.
418 As a controller regulates the distribution of the target resource to
419 the cgroup's children, enabling it creates the controller's interface
420 files in the child cgroups. In the above example, enabling "cpu" on B
421 would create the "cpu." prefixed controller interface files in C and
422 D. Likewise, disabling "memory" from B would remove the "memory."
423 prefixed controller interface files from C and D. This means that the
424 controller interface files - anything which doesn't start with
425 "cgroup." are owned by the parent rather than the cgroup itself.
431 Resources are distributed top-down and a cgroup can further distribute
432 a resource only if the resource has been distributed to it from the
433 parent. This means that all non-root "cgroup.subtree_control" files
434 can only contain controllers which are enabled in the parent's
435 "cgroup.subtree_control" file. A controller can be enabled only if
436 the parent has the controller enabled and a controller can't be
437 disabled if one or more children have it enabled.
440 No Internal Process Constraint
441 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
443 Non-root cgroups can distribute domain resources to their children
444 only when they don't have any processes of their own. In other words,
445 only domain cgroups which don't contain any processes can have domain
446 controllers enabled in their "cgroup.subtree_control" files.
448 This guarantees that, when a domain controller is looking at the part
449 of the hierarchy which has it enabled, processes are always only on
450 the leaves. This rules out situations where child cgroups compete
451 against internal processes of the parent.
453 The root cgroup is exempt from this restriction. Root contains
454 processes and anonymous resource consumption which can't be associated
455 with any other cgroups and requires special treatment from most
456 controllers. How resource consumption in the root cgroup is governed
457 is up to each controller (for more information on this topic please
458 refer to the Non-normative information section in the Controllers
461 Note that the restriction doesn't get in the way if there is no
462 enabled controller in the cgroup's "cgroup.subtree_control". This is
463 important as otherwise it wouldn't be possible to create children of a
464 populated cgroup. To control resource distribution of a cgroup, the
465 cgroup must create children and transfer all its processes to the
466 children before enabling controllers in its "cgroup.subtree_control"
476 A cgroup can be delegated in two ways. First, to a less privileged
477 user by granting write access of the directory and its "cgroup.procs",
478 "cgroup.threads" and "cgroup.subtree_control" files to the user.
479 Second, if the "nsdelegate" mount option is set, automatically to a
480 cgroup namespace on namespace creation.
482 Because the resource control interface files in a given directory
483 control the distribution of the parent's resources, the delegatee
484 shouldn't be allowed to write to them. For the first method, this is
485 achieved by not granting access to these files. For the second, the
486 kernel rejects writes to all files other than "cgroup.procs" and
487 "cgroup.subtree_control" on a namespace root from inside the
490 The end results are equivalent for both delegation types. Once
491 delegated, the user can build sub-hierarchy under the directory,
492 organize processes inside it as it sees fit and further distribute the
493 resources it received from the parent. The limits and other settings
494 of all resource controllers are hierarchical and regardless of what
495 happens in the delegated sub-hierarchy, nothing can escape the
496 resource restrictions imposed by the parent.
498 Currently, cgroup doesn't impose any restrictions on the number of
499 cgroups in or nesting depth of a delegated sub-hierarchy; however,
500 this may be limited explicitly in the future.
503 Delegation Containment
504 ~~~~~~~~~~~~~~~~~~~~~~
506 A delegated sub-hierarchy is contained in the sense that processes
507 can't be moved into or out of the sub-hierarchy by the delegatee.
509 For delegations to a less privileged user, this is achieved by
510 requiring the following conditions for a process with a non-root euid
511 to migrate a target process into a cgroup by writing its PID to the
514 - The writer must have write access to the "cgroup.procs" file.
516 - The writer must have write access to the "cgroup.procs" file of the
517 common ancestor of the source and destination cgroups.
519 The above two constraints ensure that while a delegatee may migrate
520 processes around freely in the delegated sub-hierarchy it can't pull
521 in from or push out to outside the sub-hierarchy.
523 For an example, let's assume cgroups C0 and C1 have been delegated to
524 user U0 who created C00, C01 under C0 and C10 under C1 as follows and
525 all processes under C0 and C1 belong to U0::
527 ~~~~~~~~~~~~~ - C0 - C00
530 ~~~~~~~~~~~~~ - C1 - C10
532 Let's also say U0 wants to write the PID of a process which is
533 currently in C10 into "C00/cgroup.procs". U0 has write access to the
534 file; however, the common ancestor of the source cgroup C10 and the
535 destination cgroup C00 is above the points of delegation and U0 would
536 not have write access to its "cgroup.procs" files and thus the write
537 will be denied with -EACCES.
539 For delegations to namespaces, containment is achieved by requiring
540 that both the source and destination cgroups are reachable from the
541 namespace of the process which is attempting the migration. If either
542 is not reachable, the migration is rejected with -ENOENT.
548 Organize Once and Control
549 ~~~~~~~~~~~~~~~~~~~~~~~~~
551 Migrating a process across cgroups is a relatively expensive operation
552 and stateful resources such as memory are not moved together with the
553 process. This is an explicit design decision as there often exist
554 inherent trade-offs between migration and various hot paths in terms
555 of synchronization cost.
557 As such, migrating processes across cgroups frequently as a means to
558 apply different resource restrictions is discouraged. A workload
559 should be assigned to a cgroup according to the system's logical and
560 resource structure once on start-up. Dynamic adjustments to resource
561 distribution can be made by changing controller configuration through
565 Avoid Name Collisions
566 ~~~~~~~~~~~~~~~~~~~~~
568 Interface files for a cgroup and its children cgroups occupy the same
569 directory and it is possible to create children cgroups which collide
570 with interface files.
572 All cgroup core interface files are prefixed with "cgroup." and each
573 controller's interface files are prefixed with the controller name and
574 a dot. A controller's name is composed of lower case alphabets and
575 '_'s but never begins with an '_' so it can be used as the prefix
576 character for collision avoidance. Also, interface file names won't
577 start or end with terms which are often used in categorizing workloads
578 such as job, service, slice, unit or workload.
580 cgroup doesn't do anything to prevent name collisions and it's the
581 user's responsibility to avoid them.
584 Resource Distribution Models
585 ============================
587 cgroup controllers implement several resource distribution schemes
588 depending on the resource type and expected use cases. This section
589 describes major schemes in use along with their expected behaviors.
595 A parent's resource is distributed by adding up the weights of all
596 active children and giving each the fraction matching the ratio of its
597 weight against the sum. As only children which can make use of the
598 resource at the moment participate in the distribution, this is
599 work-conserving. Due to the dynamic nature, this model is usually
600 used for stateless resources.
602 All weights are in the range [1, 10000] with the default at 100. This
603 allows symmetric multiplicative biases in both directions at fine
604 enough granularity while staying in the intuitive range.
606 As long as the weight is in range, all configuration combinations are
607 valid and there is no reason to reject configuration changes or
610 "cpu.weight" proportionally distributes CPU cycles to active children
611 and is an example of this type.
617 A child can only consume upto the configured amount of the resource.
618 Limits can be over-committed - the sum of the limits of children can
619 exceed the amount of resource available to the parent.
621 Limits are in the range [0, max] and defaults to "max", which is noop.
623 As limits can be over-committed, all configuration combinations are
624 valid and there is no reason to reject configuration changes or
627 "io.max" limits the maximum BPS and/or IOPS that a cgroup can consume
628 on an IO device and is an example of this type.
634 A cgroup is protected upto the configured amount of the resource
635 as long as the usages of all its ancestors are under their
636 protected levels. Protections can be hard guarantees or best effort
637 soft boundaries. Protections can also be over-committed in which case
638 only upto the amount available to the parent is protected among
641 Protections are in the range [0, max] and defaults to 0, which is
644 As protections can be over-committed, all configuration combinations
645 are valid and there is no reason to reject configuration changes or
648 "memory.low" implements best-effort memory protection and is an
649 example of this type.
655 A cgroup is exclusively allocated a certain amount of a finite
656 resource. Allocations can't be over-committed - the sum of the
657 allocations of children can not exceed the amount of resource
658 available to the parent.
660 Allocations are in the range [0, max] and defaults to 0, which is no
663 As allocations can't be over-committed, some configuration
664 combinations are invalid and should be rejected. Also, if the
665 resource is mandatory for execution of processes, process migrations
668 "cpu.rt.max" hard-allocates realtime slices and is an example of this
678 All interface files should be in one of the following formats whenever
681 New-line separated values
682 (when only one value can be written at once)
688 Space separated values
689 (when read-only or multiple values can be written at once)
701 KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01...
702 KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11...
705 For a writable file, the format for writing should generally match
706 reading; however, controllers may allow omitting later fields or
707 implement restricted shortcuts for most common use cases.
709 For both flat and nested keyed files, only the values for a single key
710 can be written at a time. For nested keyed files, the sub key pairs
711 may be specified in any order and not all pairs have to be specified.
717 - Settings for a single feature should be contained in a single file.
719 - The root cgroup should be exempt from resource control and thus
720 shouldn't have resource control interface files.
722 - The default time unit is microseconds. If a different unit is ever
723 used, an explicit unit suffix must be present.
725 - A parts-per quantity should use a percentage decimal with at least
726 two digit fractional part - e.g. 13.40.
728 - If a controller implements weight based resource distribution, its
729 interface file should be named "weight" and have the range [1,
730 10000] with 100 as the default. The values are chosen to allow
731 enough and symmetric bias in both directions while keeping it
732 intuitive (the default is 100%).
734 - If a controller implements an absolute resource guarantee and/or
735 limit, the interface files should be named "min" and "max"
736 respectively. If a controller implements best effort resource
737 guarantee and/or limit, the interface files should be named "low"
738 and "high" respectively.
740 In the above four control files, the special token "max" should be
741 used to represent upward infinity for both reading and writing.
743 - If a setting has a configurable default value and keyed specific
744 overrides, the default entry should be keyed with "default" and
745 appear as the first entry in the file.
747 The default value can be updated by writing either "default $VAL" or
750 When writing to update a specific override, "default" can be used as
751 the value to indicate removal of the override. Override entries
752 with "default" as the value must not appear when read.
754 For example, a setting which is keyed by major:minor device numbers
755 with integer values may look like the following::
757 # cat cgroup-example-interface-file
761 The default value can be updated by::
763 # echo 125 > cgroup-example-interface-file
767 # echo "default 125" > cgroup-example-interface-file
769 An override can be set by::
771 # echo "8:16 170" > cgroup-example-interface-file
775 # echo "8:0 default" > cgroup-example-interface-file
776 # cat cgroup-example-interface-file
780 - For events which are not very high frequency, an interface file
781 "events" should be created which lists event key value pairs.
782 Whenever a notifiable event happens, file modified event should be
783 generated on the file.
789 All cgroup core files are prefixed with "cgroup."
792 A read-write single value file which exists on non-root
795 When read, it indicates the current type of the cgroup, which
796 can be one of the following values.
798 - "domain" : A normal valid domain cgroup.
800 - "domain threaded" : A threaded domain cgroup which is
801 serving as the root of a threaded subtree.
803 - "domain invalid" : A cgroup which is in an invalid state.
804 It can't be populated or have controllers enabled. It may
805 be allowed to become a threaded cgroup.
807 - "threaded" : A threaded cgroup which is a member of a
810 A cgroup can be turned into a threaded cgroup by writing
811 "threaded" to this file.
814 A read-write new-line separated values file which exists on
817 When read, it lists the PIDs of all processes which belong to
818 the cgroup one-per-line. The PIDs are not ordered and the
819 same PID may show up more than once if the process got moved
820 to another cgroup and then back or the PID got recycled while
823 A PID can be written to migrate the process associated with
824 the PID to the cgroup. The writer should match all of the
825 following conditions.
827 - It must have write access to the "cgroup.procs" file.
829 - It must have write access to the "cgroup.procs" file of the
830 common ancestor of the source and destination cgroups.
832 When delegating a sub-hierarchy, write access to this file
833 should be granted along with the containing directory.
835 In a threaded cgroup, reading this file fails with EOPNOTSUPP
836 as all the processes belong to the thread root. Writing is
837 supported and moves every thread of the process to the cgroup.
840 A read-write new-line separated values file which exists on
843 When read, it lists the TIDs of all threads which belong to
844 the cgroup one-per-line. The TIDs are not ordered and the
845 same TID may show up more than once if the thread got moved to
846 another cgroup and then back or the TID got recycled while
849 A TID can be written to migrate the thread associated with the
850 TID to the cgroup. The writer should match all of the
851 following conditions.
853 - It must have write access to the "cgroup.threads" file.
855 - The cgroup that the thread is currently in must be in the
856 same resource domain as the destination cgroup.
858 - It must have write access to the "cgroup.procs" file of the
859 common ancestor of the source and destination cgroups.
861 When delegating a sub-hierarchy, write access to this file
862 should be granted along with the containing directory.
865 A read-only space separated values file which exists on all
868 It shows space separated list of all controllers available to
869 the cgroup. The controllers are not ordered.
871 cgroup.subtree_control
872 A read-write space separated values file which exists on all
873 cgroups. Starts out empty.
875 When read, it shows space separated list of the controllers
876 which are enabled to control resource distribution from the
877 cgroup to its children.
879 Space separated list of controllers prefixed with '+' or '-'
880 can be written to enable or disable controllers. A controller
881 name prefixed with '+' enables the controller and '-'
882 disables. If a controller appears more than once on the list,
883 the last one is effective. When multiple enable and disable
884 operations are specified, either all succeed or all fail.
887 A read-only flat-keyed file which exists on non-root cgroups.
888 The following entries are defined. Unless specified
889 otherwise, a value change in this file generates a file
893 1 if the cgroup or its descendants contains any live
894 processes; otherwise, 0.
896 1 if the cgroup is frozen; otherwise, 0.
898 cgroup.max.descendants
899 A read-write single value files. The default is "max".
901 Maximum allowed number of descent cgroups.
902 If the actual number of descendants is equal or larger,
903 an attempt to create a new cgroup in the hierarchy will fail.
906 A read-write single value files. The default is "max".
908 Maximum allowed descent depth below the current cgroup.
909 If the actual descent depth is equal or larger,
910 an attempt to create a new child cgroup will fail.
913 A read-only flat-keyed file with the following entries:
916 Total number of visible descendant cgroups.
919 Total number of dying descendant cgroups. A cgroup becomes
920 dying after being deleted by a user. The cgroup will remain
921 in dying state for some time undefined time (which can depend
922 on system load) before being completely destroyed.
924 A process can't enter a dying cgroup under any circumstances,
925 a dying cgroup can't revive.
927 A dying cgroup can consume system resources not exceeding
928 limits, which were active at the moment of cgroup deletion.
931 A read-write single value file which exists on non-root cgroups.
932 Allowed values are "0" and "1". The default is "0".
934 Writing "1" to the file causes freezing of the cgroup and all
935 descendant cgroups. This means that all belonging processes will
936 be stopped and will not run until the cgroup will be explicitly
937 unfrozen. Freezing of the cgroup may take some time; when this action
938 is completed, the "frozen" value in the cgroup.events control file
939 will be updated to "1" and the corresponding notification will be
942 A cgroup can be frozen either by its own settings, or by settings
943 of any ancestor cgroups. If any of ancestor cgroups is frozen, the
944 cgroup will remain frozen.
946 Processes in the frozen cgroup can be killed by a fatal signal.
947 They also can enter and leave a frozen cgroup: either by an explicit
948 move by a user, or if freezing of the cgroup races with fork().
949 If a process is moved to a frozen cgroup, it stops. If a process is
950 moved out of a frozen cgroup, it becomes running.
952 Frozen status of a cgroup doesn't affect any cgroup tree operations:
953 it's possible to delete a frozen (and empty) cgroup, as well as
954 create new sub-cgroups.
957 A write-only single value file which exists in non-root cgroups.
958 The only allowed value is "1".
960 Writing "1" to the file causes the cgroup and all descendant cgroups to
961 be killed. This means that all processes located in the affected cgroup
962 tree will be killed via SIGKILL.
964 Killing a cgroup tree will deal with concurrent forks appropriately and
965 is protected against migrations.
967 In a threaded cgroup, writing this file fails with EOPNOTSUPP as
968 killing cgroups is a process directed operation, i.e. it affects
969 the whole thread-group.
979 The "cpu" controllers regulates distribution of CPU cycles. This
980 controller implements weight and absolute bandwidth limit models for
981 normal scheduling policy and absolute bandwidth allocation model for
982 realtime scheduling policy.
984 In all the above models, cycles distribution is defined only on a temporal
985 base and it does not account for the frequency at which tasks are executed.
986 The (optional) utilization clamping support allows to hint the schedutil
987 cpufreq governor about the minimum desired frequency which should always be
988 provided by a CPU, as well as the maximum desired frequency, which should not
989 be exceeded by a CPU.
991 WARNING: cgroup2 doesn't yet support control of realtime processes and
992 the cpu controller can only be enabled when all RT processes are in
993 the root cgroup. Be aware that system management software may already
994 have placed RT processes into nonroot cgroups during the system boot
995 process, and these processes may need to be moved to the root cgroup
996 before the cpu controller can be enabled.
1002 All time durations are in microseconds.
1005 A read-only flat-keyed file.
1006 This file exists whether the controller is enabled or not.
1008 It always reports the following three stats:
1014 and the following three when the controller is enabled:
1023 A read-write single value file which exists on non-root
1024 cgroups. The default is "100".
1026 The weight in the range [1, 10000].
1029 A read-write single value file which exists on non-root
1030 cgroups. The default is "0".
1032 The nice value is in the range [-20, 19].
1034 This interface file is an alternative interface for
1035 "cpu.weight" and allows reading and setting weight using the
1036 same values used by nice(2). Because the range is smaller and
1037 granularity is coarser for the nice values, the read value is
1038 the closest approximation of the current weight.
1041 A read-write two value file which exists on non-root cgroups.
1042 The default is "max 100000".
1044 The maximum bandwidth limit. It's in the following format::
1048 which indicates that the group may consume upto $MAX in each
1049 $PERIOD duration. "max" for $MAX indicates no limit. If only
1050 one number is written, $MAX is updated.
1053 A read-write single value file which exists on non-root
1054 cgroups. The default is "0".
1056 The burst in the range [0, $MAX].
1059 A read-write nested-keyed file.
1061 Shows pressure stall information for CPU. See
1062 :ref:`Documentation/accounting/psi.rst <psi>` for details.
1065 A read-write single value file which exists on non-root cgroups.
1066 The default is "0", i.e. no utilization boosting.
1068 The requested minimum utilization (protection) as a percentage
1069 rational number, e.g. 12.34 for 12.34%.
1071 This interface allows reading and setting minimum utilization clamp
1072 values similar to the sched_setattr(2). This minimum utilization
1073 value is used to clamp the task specific minimum utilization clamp.
1075 The requested minimum utilization (protection) is always capped by
1076 the current value for the maximum utilization (limit), i.e.
1080 A read-write single value file which exists on non-root cgroups.
1081 The default is "max". i.e. no utilization capping
1083 The requested maximum utilization (limit) as a percentage rational
1084 number, e.g. 98.76 for 98.76%.
1086 This interface allows reading and setting maximum utilization clamp
1087 values similar to the sched_setattr(2). This maximum utilization
1088 value is used to clamp the task specific maximum utilization clamp.
1095 The "memory" controller regulates distribution of memory. Memory is
1096 stateful and implements both limit and protection models. Due to the
1097 intertwining between memory usage and reclaim pressure and the
1098 stateful nature of memory, the distribution model is relatively
1101 While not completely water-tight, all major memory usages by a given
1102 cgroup are tracked so that the total memory consumption can be
1103 accounted and controlled to a reasonable extent. Currently, the
1104 following types of memory usages are tracked.
1106 - Userland memory - page cache and anonymous memory.
1108 - Kernel data structures such as dentries and inodes.
1110 - TCP socket buffers.
1112 The above list may expand in the future for better coverage.
1115 Memory Interface Files
1116 ~~~~~~~~~~~~~~~~~~~~~~
1118 All memory amounts are in bytes. If a value which is not aligned to
1119 PAGE_SIZE is written, the value may be rounded up to the closest
1120 PAGE_SIZE multiple when read back.
1123 A read-only single value file which exists on non-root
1126 The total amount of memory currently being used by the cgroup
1127 and its descendants.
1130 A read-write single value file which exists on non-root
1131 cgroups. The default is "0".
1133 Hard memory protection. If the memory usage of a cgroup
1134 is within its effective min boundary, the cgroup's memory
1135 won't be reclaimed under any conditions. If there is no
1136 unprotected reclaimable memory available, OOM killer
1137 is invoked. Above the effective min boundary (or
1138 effective low boundary if it is higher), pages are reclaimed
1139 proportionally to the overage, reducing reclaim pressure for
1142 Effective min boundary is limited by memory.min values of
1143 all ancestor cgroups. If there is memory.min overcommitment
1144 (child cgroup or cgroups are requiring more protected memory
1145 than parent will allow), then each child cgroup will get
1146 the part of parent's protection proportional to its
1147 actual memory usage below memory.min.
1149 Putting more memory than generally available under this
1150 protection is discouraged and may lead to constant OOMs.
1152 If a memory cgroup is not populated with processes,
1153 its memory.min is ignored.
1156 A read-write single value file which exists on non-root
1157 cgroups. The default is "0".
1159 Best-effort memory protection. If the memory usage of a
1160 cgroup is within its effective low boundary, the cgroup's
1161 memory won't be reclaimed unless there is no reclaimable
1162 memory available in unprotected cgroups.
1163 Above the effective low boundary (or
1164 effective min boundary if it is higher), pages are reclaimed
1165 proportionally to the overage, reducing reclaim pressure for
1168 Effective low boundary is limited by memory.low values of
1169 all ancestor cgroups. If there is memory.low overcommitment
1170 (child cgroup or cgroups are requiring more protected memory
1171 than parent will allow), then each child cgroup will get
1172 the part of parent's protection proportional to its
1173 actual memory usage below memory.low.
1175 Putting more memory than generally available under this
1176 protection is discouraged.
1179 A read-write single value file which exists on non-root
1180 cgroups. The default is "max".
1182 Memory usage throttle limit. This is the main mechanism to
1183 control memory usage of a cgroup. If a cgroup's usage goes
1184 over the high boundary, the processes of the cgroup are
1185 throttled and put under heavy reclaim pressure.
1187 Going over the high limit never invokes the OOM killer and
1188 under extreme conditions the limit may be breached.
1191 A read-write single value file which exists on non-root
1192 cgroups. The default is "max".
1194 Memory usage hard limit. This is the final protection
1195 mechanism. If a cgroup's memory usage reaches this limit and
1196 can't be reduced, the OOM killer is invoked in the cgroup.
1197 Under certain circumstances, the usage may go over the limit
1200 In default configuration regular 0-order allocations always
1201 succeed unless OOM killer chooses current task as a victim.
1203 Some kinds of allocations don't invoke the OOM killer.
1204 Caller could retry them differently, return into userspace
1205 as -ENOMEM or silently ignore in cases like disk readahead.
1207 This is the ultimate protection mechanism. As long as the
1208 high limit is used and monitored properly, this limit's
1209 utility is limited to providing the final safety net.
1212 A read-write single value file which exists on non-root
1213 cgroups. The default value is "0".
1215 Determines whether the cgroup should be treated as
1216 an indivisible workload by the OOM killer. If set,
1217 all tasks belonging to the cgroup or to its descendants
1218 (if the memory cgroup is not a leaf cgroup) are killed
1219 together or not at all. This can be used to avoid
1220 partial kills to guarantee workload integrity.
1222 Tasks with the OOM protection (oom_score_adj set to -1000)
1223 are treated as an exception and are never killed.
1225 If the OOM killer is invoked in a cgroup, it's not going
1226 to kill any tasks outside of this cgroup, regardless
1227 memory.oom.group values of ancestor cgroups.
1230 A read-only flat-keyed file which exists on non-root cgroups.
1231 The following entries are defined. Unless specified
1232 otherwise, a value change in this file generates a file
1235 Note that all fields in this file are hierarchical and the
1236 file modified event can be generated due to an event down the
1237 hierarchy. For the local events at the cgroup level see
1238 memory.events.local.
1241 The number of times the cgroup is reclaimed due to
1242 high memory pressure even though its usage is under
1243 the low boundary. This usually indicates that the low
1244 boundary is over-committed.
1247 The number of times processes of the cgroup are
1248 throttled and routed to perform direct memory reclaim
1249 because the high memory boundary was exceeded. For a
1250 cgroup whose memory usage is capped by the high limit
1251 rather than global memory pressure, this event's
1252 occurrences are expected.
1255 The number of times the cgroup's memory usage was
1256 about to go over the max boundary. If direct reclaim
1257 fails to bring it down, the cgroup goes to OOM state.
1260 The number of time the cgroup's memory usage was
1261 reached the limit and allocation was about to fail.
1263 This event is not raised if the OOM killer is not
1264 considered as an option, e.g. for failed high-order
1265 allocations or if caller asked to not retry attempts.
1268 The number of processes belonging to this cgroup
1269 killed by any kind of OOM killer.
1272 Similar to memory.events but the fields in the file are local
1273 to the cgroup i.e. not hierarchical. The file modified event
1274 generated on this file reflects only the local events.
1277 A read-only flat-keyed file which exists on non-root cgroups.
1279 This breaks down the cgroup's memory footprint into different
1280 types of memory, type-specific details, and other information
1281 on the state and past events of the memory management system.
1283 All memory amounts are in bytes.
1285 The entries are ordered to be human readable, and new entries
1286 can show up in the middle. Don't rely on items remaining in a
1287 fixed position; use the keys to look up specific values!
1289 If the entry has no per-node counter (or not show in the
1290 memory.numa_stat). We use 'npn' (non-per-node) as the tag
1291 to indicate that it will not show in the memory.numa_stat.
1294 Amount of memory used in anonymous mappings such as
1295 brk(), sbrk(), and mmap(MAP_ANONYMOUS)
1298 Amount of memory used to cache filesystem data,
1299 including tmpfs and shared memory.
1302 Amount of memory allocated to kernel stacks.
1305 Amount of memory allocated for page tables.
1308 Amount of memory used for storing per-cpu kernel
1312 Amount of memory used in network transmission buffers
1315 Amount of cached filesystem data that is swap-backed,
1316 such as tmpfs, shm segments, shared anonymous mmap()s
1319 Amount of cached filesystem data mapped with mmap()
1322 Amount of cached filesystem data that was modified but
1323 not yet written back to disk
1326 Amount of cached filesystem data that was modified and
1327 is currently being written back to disk
1330 Amount of swap cached in memory. The swapcache is accounted
1331 against both memory and swap usage.
1334 Amount of memory used in anonymous mappings backed by
1335 transparent hugepages
1338 Amount of cached filesystem data backed by transparent
1342 Amount of shm, tmpfs, shared anonymous mmap()s backed by
1343 transparent hugepages
1345 inactive_anon, active_anon, inactive_file, active_file, unevictable
1346 Amount of memory, swap-backed and filesystem-backed,
1347 on the internal memory management lists used by the
1348 page reclaim algorithm.
1350 As these represent internal list state (eg. shmem pages are on anon
1351 memory management lists), inactive_foo + active_foo may not be equal to
1352 the value for the foo counter, since the foo counter is type-based, not
1356 Part of "slab" that might be reclaimed, such as
1357 dentries and inodes.
1360 Part of "slab" that cannot be reclaimed on memory
1364 Amount of memory used for storing in-kernel data
1367 workingset_refault_anon
1368 Number of refaults of previously evicted anonymous pages.
1370 workingset_refault_file
1371 Number of refaults of previously evicted file pages.
1373 workingset_activate_anon
1374 Number of refaulted anonymous pages that were immediately
1377 workingset_activate_file
1378 Number of refaulted file pages that were immediately activated.
1380 workingset_restore_anon
1381 Number of restored anonymous pages which have been detected as
1382 an active workingset before they got reclaimed.
1384 workingset_restore_file
1385 Number of restored file pages which have been detected as an
1386 active workingset before they got reclaimed.
1388 workingset_nodereclaim
1389 Number of times a shadow node has been reclaimed
1392 Total number of page faults incurred
1395 Number of major page faults incurred
1398 Amount of scanned pages (in an active LRU list)
1401 Amount of scanned pages (in an inactive LRU list)
1404 Amount of reclaimed pages
1407 Amount of pages moved to the active LRU list
1410 Amount of pages moved to the inactive LRU list
1413 Amount of pages postponed to be freed under memory pressure
1416 Amount of reclaimed lazyfree pages
1418 thp_fault_alloc (npn)
1419 Number of transparent hugepages which were allocated to satisfy
1420 a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE
1423 thp_collapse_alloc (npn)
1424 Number of transparent hugepages which were allocated to allow
1425 collapsing an existing range of pages. This counter is not
1426 present when CONFIG_TRANSPARENT_HUGEPAGE is not set.
1429 A read-only nested-keyed file which exists on non-root cgroups.
1431 This breaks down the cgroup's memory footprint into different
1432 types of memory, type-specific details, and other information
1433 per node on the state of the memory management system.
1435 This is useful for providing visibility into the NUMA locality
1436 information within an memcg since the pages are allowed to be
1437 allocated from any physical node. One of the use case is evaluating
1438 application performance by combining this information with the
1439 application's CPU allocation.
1441 All memory amounts are in bytes.
1443 The output format of memory.numa_stat is::
1445 type N0=<bytes in node 0> N1=<bytes in node 1> ...
1447 The entries are ordered to be human readable, and new entries
1448 can show up in the middle. Don't rely on items remaining in a
1449 fixed position; use the keys to look up specific values!
1451 The entries can refer to the memory.stat.
1454 A read-only single value file which exists on non-root
1457 The total amount of swap currently being used by the cgroup
1458 and its descendants.
1461 A read-write single value file which exists on non-root
1462 cgroups. The default is "max".
1464 Swap usage throttle limit. If a cgroup's swap usage exceeds
1465 this limit, all its further allocations will be throttled to
1466 allow userspace to implement custom out-of-memory procedures.
1468 This limit marks a point of no return for the cgroup. It is NOT
1469 designed to manage the amount of swapping a workload does
1470 during regular operation. Compare to memory.swap.max, which
1471 prohibits swapping past a set amount, but lets the cgroup
1472 continue unimpeded as long as other memory can be reclaimed.
1474 Healthy workloads are not expected to reach this limit.
1477 A read-write single value file which exists on non-root
1478 cgroups. The default is "max".
1480 Swap usage hard limit. If a cgroup's swap usage reaches this
1481 limit, anonymous memory of the cgroup will not be swapped out.
1484 A read-only flat-keyed file which exists on non-root cgroups.
1485 The following entries are defined. Unless specified
1486 otherwise, a value change in this file generates a file
1490 The number of times the cgroup's swap usage was over
1494 The number of times the cgroup's swap usage was about
1495 to go over the max boundary and swap allocation
1499 The number of times swap allocation failed either
1500 because of running out of swap system-wide or max
1503 When reduced under the current usage, the existing swap
1504 entries are reclaimed gradually and the swap usage may stay
1505 higher than the limit for an extended period of time. This
1506 reduces the impact on the workload and memory management.
1509 A read-only nested-keyed file.
1511 Shows pressure stall information for memory. See
1512 :ref:`Documentation/accounting/psi.rst <psi>` for details.
1518 "memory.high" is the main mechanism to control memory usage.
1519 Over-committing on high limit (sum of high limits > available memory)
1520 and letting global memory pressure to distribute memory according to
1521 usage is a viable strategy.
1523 Because breach of the high limit doesn't trigger the OOM killer but
1524 throttles the offending cgroup, a management agent has ample
1525 opportunities to monitor and take appropriate actions such as granting
1526 more memory or terminating the workload.
1528 Determining whether a cgroup has enough memory is not trivial as
1529 memory usage doesn't indicate whether the workload can benefit from
1530 more memory. For example, a workload which writes data received from
1531 network to a file can use all available memory but can also operate as
1532 performant with a small amount of memory. A measure of memory
1533 pressure - how much the workload is being impacted due to lack of
1534 memory - is necessary to determine whether a workload needs more
1535 memory; unfortunately, memory pressure monitoring mechanism isn't
1542 A memory area is charged to the cgroup which instantiated it and stays
1543 charged to the cgroup until the area is released. Migrating a process
1544 to a different cgroup doesn't move the memory usages that it
1545 instantiated while in the previous cgroup to the new cgroup.
1547 A memory area may be used by processes belonging to different cgroups.
1548 To which cgroup the area will be charged is in-deterministic; however,
1549 over time, the memory area is likely to end up in a cgroup which has
1550 enough memory allowance to avoid high reclaim pressure.
1552 If a cgroup sweeps a considerable amount of memory which is expected
1553 to be accessed repeatedly by other cgroups, it may make sense to use
1554 POSIX_FADV_DONTNEED to relinquish the ownership of memory areas
1555 belonging to the affected files to ensure correct memory ownership.
1561 The "io" controller regulates the distribution of IO resources. This
1562 controller implements both weight based and absolute bandwidth or IOPS
1563 limit distribution; however, weight based distribution is available
1564 only if cfq-iosched is in use and neither scheme is available for
1572 A read-only nested-keyed file.
1574 Lines are keyed by $MAJ:$MIN device numbers and not ordered.
1575 The following nested keys are defined.
1577 ====== =====================
1579 wbytes Bytes written
1580 rios Number of read IOs
1581 wios Number of write IOs
1582 dbytes Bytes discarded
1583 dios Number of discard IOs
1584 ====== =====================
1586 An example read output follows::
1588 8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
1589 8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
1592 A read-write nested-keyed file which exists only on the root
1595 This file configures the Quality of Service of the IO cost
1596 model based controller (CONFIG_BLK_CGROUP_IOCOST) which
1597 currently implements "io.weight" proportional control. Lines
1598 are keyed by $MAJ:$MIN device numbers and not ordered. The
1599 line for a given device is populated on the first write for
1600 the device on "io.cost.qos" or "io.cost.model". The following
1601 nested keys are defined.
1603 ====== =====================================
1604 enable Weight-based control enable
1605 ctrl "auto" or "user"
1606 rpct Read latency percentile [0, 100]
1607 rlat Read latency threshold
1608 wpct Write latency percentile [0, 100]
1609 wlat Write latency threshold
1610 min Minimum scaling percentage [1, 10000]
1611 max Maximum scaling percentage [1, 10000]
1612 ====== =====================================
1614 The controller is disabled by default and can be enabled by
1615 setting "enable" to 1. "rpct" and "wpct" parameters default
1616 to zero and the controller uses internal device saturation
1617 state to adjust the overall IO rate between "min" and "max".
1619 When a better control quality is needed, latency QoS
1620 parameters can be configured. For example::
1622 8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
1624 shows that on sdb, the controller is enabled, will consider
1625 the device saturated if the 95th percentile of read completion
1626 latencies is above 75ms or write 150ms, and adjust the overall
1627 IO issue rate between 50% and 150% accordingly.
1629 The lower the saturation point, the better the latency QoS at
1630 the cost of aggregate bandwidth. The narrower the allowed
1631 adjustment range between "min" and "max", the more conformant
1632 to the cost model the IO behavior. Note that the IO issue
1633 base rate may be far off from 100% and setting "min" and "max"
1634 blindly can lead to a significant loss of device capacity or
1635 control quality. "min" and "max" are useful for regulating
1636 devices which show wide temporary behavior changes - e.g. a
1637 ssd which accepts writes at the line speed for a while and
1638 then completely stalls for multiple seconds.
1640 When "ctrl" is "auto", the parameters are controlled by the
1641 kernel and may change automatically. Setting "ctrl" to "user"
1642 or setting any of the percentile and latency parameters puts
1643 it into "user" mode and disables the automatic changes. The
1644 automatic mode can be restored by setting "ctrl" to "auto".
1647 A read-write nested-keyed file which exists only on the root
1650 This file configures the cost model of the IO cost model based
1651 controller (CONFIG_BLK_CGROUP_IOCOST) which currently
1652 implements "io.weight" proportional control. Lines are keyed
1653 by $MAJ:$MIN device numbers and not ordered. The line for a
1654 given device is populated on the first write for the device on
1655 "io.cost.qos" or "io.cost.model". The following nested keys
1658 ===== ================================
1659 ctrl "auto" or "user"
1660 model The cost model in use - "linear"
1661 ===== ================================
1663 When "ctrl" is "auto", the kernel may change all parameters
1664 dynamically. When "ctrl" is set to "user" or any other
1665 parameters are written to, "ctrl" become "user" and the
1666 automatic changes are disabled.
1668 When "model" is "linear", the following model parameters are
1671 ============= ========================================
1672 [r|w]bps The maximum sequential IO throughput
1673 [r|w]seqiops The maximum 4k sequential IOs per second
1674 [r|w]randiops The maximum 4k random IOs per second
1675 ============= ========================================
1677 From the above, the builtin linear model determines the base
1678 costs of a sequential and random IO and the cost coefficient
1679 for the IO size. While simple, this model can cover most
1680 common device classes acceptably.
1682 The IO cost model isn't expected to be accurate in absolute
1683 sense and is scaled to the device behavior dynamically.
1685 If needed, tools/cgroup/iocost_coef_gen.py can be used to
1686 generate device-specific coefficients.
1689 A read-write flat-keyed file which exists on non-root cgroups.
1690 The default is "default 100".
1692 The first line is the default weight applied to devices
1693 without specific override. The rest are overrides keyed by
1694 $MAJ:$MIN device numbers and not ordered. The weights are in
1695 the range [1, 10000] and specifies the relative amount IO time
1696 the cgroup can use in relation to its siblings.
1698 The default weight can be updated by writing either "default
1699 $WEIGHT" or simply "$WEIGHT". Overrides can be set by writing
1700 "$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default".
1702 An example read output follows::
1709 A read-write nested-keyed file which exists on non-root
1712 BPS and IOPS based IO limit. Lines are keyed by $MAJ:$MIN
1713 device numbers and not ordered. The following nested keys are
1716 ===== ==================================
1717 rbps Max read bytes per second
1718 wbps Max write bytes per second
1719 riops Max read IO operations per second
1720 wiops Max write IO operations per second
1721 ===== ==================================
1723 When writing, any number of nested key-value pairs can be
1724 specified in any order. "max" can be specified as the value
1725 to remove a specific limit. If the same key is specified
1726 multiple times, the outcome is undefined.
1728 BPS and IOPS are measured in each IO direction and IOs are
1729 delayed if limit is reached. Temporary bursts are allowed.
1731 Setting read limit at 2M BPS and write at 120 IOPS for 8:16::
1733 echo "8:16 rbps=2097152 wiops=120" > io.max
1735 Reading returns the following::
1737 8:16 rbps=2097152 wbps=max riops=max wiops=120
1739 Write IOPS limit can be removed by writing the following::
1741 echo "8:16 wiops=max" > io.max
1743 Reading now returns the following::
1745 8:16 rbps=2097152 wbps=max riops=max wiops=max
1748 A read-only nested-keyed file.
1750 Shows pressure stall information for IO. See
1751 :ref:`Documentation/accounting/psi.rst <psi>` for details.
1757 Page cache is dirtied through buffered writes and shared mmaps and
1758 written asynchronously to the backing filesystem by the writeback
1759 mechanism. Writeback sits between the memory and IO domains and
1760 regulates the proportion of dirty memory by balancing dirtying and
1763 The io controller, in conjunction with the memory controller,
1764 implements control of page cache writeback IOs. The memory controller
1765 defines the memory domain that dirty memory ratio is calculated and
1766 maintained for and the io controller defines the io domain which
1767 writes out dirty pages for the memory domain. Both system-wide and
1768 per-cgroup dirty memory states are examined and the more restrictive
1769 of the two is enforced.
1771 cgroup writeback requires explicit support from the underlying
1772 filesystem. Currently, cgroup writeback is implemented on ext2, ext4,
1773 btrfs, f2fs, and xfs. On other filesystems, all writeback IOs are
1774 attributed to the root cgroup.
1776 There are inherent differences in memory and writeback management
1777 which affects how cgroup ownership is tracked. Memory is tracked per
1778 page while writeback per inode. For the purpose of writeback, an
1779 inode is assigned to a cgroup and all IO requests to write dirty pages
1780 from the inode are attributed to that cgroup.
1782 As cgroup ownership for memory is tracked per page, there can be pages
1783 which are associated with different cgroups than the one the inode is
1784 associated with. These are called foreign pages. The writeback
1785 constantly keeps track of foreign pages and, if a particular foreign
1786 cgroup becomes the majority over a certain period of time, switches
1787 the ownership of the inode to that cgroup.
1789 While this model is enough for most use cases where a given inode is
1790 mostly dirtied by a single cgroup even when the main writing cgroup
1791 changes over time, use cases where multiple cgroups write to a single
1792 inode simultaneously are not supported well. In such circumstances, a
1793 significant portion of IOs are likely to be attributed incorrectly.
1794 As memory controller assigns page ownership on the first use and
1795 doesn't update it until the page is released, even if writeback
1796 strictly follows page ownership, multiple cgroups dirtying overlapping
1797 areas wouldn't work as expected. It's recommended to avoid such usage
1800 The sysctl knobs which affect writeback behavior are applied to cgroup
1801 writeback as follows.
1803 vm.dirty_background_ratio, vm.dirty_ratio
1804 These ratios apply the same to cgroup writeback with the
1805 amount of available memory capped by limits imposed by the
1806 memory controller and system-wide clean memory.
1808 vm.dirty_background_bytes, vm.dirty_bytes
1809 For cgroup writeback, this is calculated into ratio against
1810 total available memory and applied the same way as
1811 vm.dirty[_background]_ratio.
1817 This is a cgroup v2 controller for IO workload protection. You provide a group
1818 with a latency target, and if the average latency exceeds that target the
1819 controller will throttle any peers that have a lower latency target than the
1822 The limits are only applied at the peer level in the hierarchy. This means that
1823 in the diagram below, only groups A, B, and C will influence each other, and
1824 groups D and F will influence each other. Group G will influence nobody::
1833 So the ideal way to configure this is to set io.latency in groups A, B, and C.
1834 Generally you do not want to set a value lower than the latency your device
1835 supports. Experiment to find the value that works best for your workload.
1836 Start at higher than the expected latency for your device and watch the
1837 avg_lat value in io.stat for your workload group to get an idea of the
1838 latency you see during normal operation. Use the avg_lat value as a basis for
1839 your real setting, setting at 10-15% higher than the value in io.stat.
1841 How IO Latency Throttling Works
1842 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1844 io.latency is work conserving; so as long as everybody is meeting their latency
1845 target the controller doesn't do anything. Once a group starts missing its
1846 target it begins throttling any peer group that has a higher target than itself.
1847 This throttling takes 2 forms:
1849 - Queue depth throttling. This is the number of outstanding IO's a group is
1850 allowed to have. We will clamp down relatively quickly, starting at no limit
1851 and going all the way down to 1 IO at a time.
1853 - Artificial delay induction. There are certain types of IO that cannot be
1854 throttled without possibly adversely affecting higher priority groups. This
1855 includes swapping and metadata IO. These types of IO are allowed to occur
1856 normally, however they are "charged" to the originating group. If the
1857 originating group is being throttled you will see the use_delay and delay
1858 fields in io.stat increase. The delay value is how many microseconds that are
1859 being added to any process that runs in this group. Because this number can
1860 grow quite large if there is a lot of swapping or metadata IO occurring we
1861 limit the individual delay events to 1 second at a time.
1863 Once the victimized group starts meeting its latency target again it will start
1864 unthrottling any peer groups that were throttled previously. If the victimized
1865 group simply stops doing IO the global counter will unthrottle appropriately.
1867 IO Latency Interface Files
1868 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1871 This takes a similar format as the other controllers.
1873 "MAJOR:MINOR target=<target time in microseconds"
1876 If the controller is enabled you will see extra stats in io.stat in
1877 addition to the normal ones.
1880 This is the current queue depth for the group.
1883 This is an exponential moving average with a decay rate of 1/exp
1884 bound by the sampling interval. The decay rate interval can be
1885 calculated by multiplying the win value in io.stat by the
1886 corresponding number of samples based on the win value.
1889 The sampling window size in milliseconds. This is the minimum
1890 duration of time between evaluation events. Windows only elapse
1891 with IO activity. Idle periods extend the most recent window.
1896 A single attribute controls the behavior of the I/O priority cgroup policy,
1897 namely the blkio.prio.class attribute. The following values are accepted for
1901 Do not modify the I/O priority class.
1904 For requests that do not have an I/O priority class (NONE),
1905 change the I/O priority class into RT. Do not modify
1906 the I/O priority class of other requests.
1909 For requests that do not have an I/O priority class or that have I/O
1910 priority class RT, change it into BE. Do not modify the I/O priority
1911 class of requests that have priority class IDLE.
1914 Change the I/O priority class of all requests into IDLE, the lowest
1917 The following numerical values are associated with the I/O priority policies:
1929 The numerical value that corresponds to each I/O priority class is as follows:
1931 +-------------------------------+---+
1932 | IOPRIO_CLASS_NONE | 0 |
1933 +-------------------------------+---+
1934 | IOPRIO_CLASS_RT (real-time) | 1 |
1935 +-------------------------------+---+
1936 | IOPRIO_CLASS_BE (best effort) | 2 |
1937 +-------------------------------+---+
1938 | IOPRIO_CLASS_IDLE | 3 |
1939 +-------------------------------+---+
1941 The algorithm to set the I/O priority class for a request is as follows:
1943 - Translate the I/O priority class policy into a number.
1944 - Change the request I/O priority class into the maximum of the I/O priority
1945 class policy number and the numerical I/O priority class.
1950 The process number controller is used to allow a cgroup to stop any
1951 new tasks from being fork()'d or clone()'d after a specified limit is
1954 The number of tasks in a cgroup can be exhausted in ways which other
1955 controllers cannot prevent, thus warranting its own controller. For
1956 example, a fork bomb is likely to exhaust the number of tasks before
1957 hitting memory restrictions.
1959 Note that PIDs used in this controller refer to TIDs, process IDs as
1967 A read-write single value file which exists on non-root
1968 cgroups. The default is "max".
1970 Hard limit of number of processes.
1973 A read-only single value file which exists on all cgroups.
1975 The number of processes currently in the cgroup and its
1978 Organisational operations are not blocked by cgroup policies, so it is
1979 possible to have pids.current > pids.max. This can be done by either
1980 setting the limit to be smaller than pids.current, or attaching enough
1981 processes to the cgroup such that pids.current is larger than
1982 pids.max. However, it is not possible to violate a cgroup PID policy
1983 through fork() or clone(). These will return -EAGAIN if the creation
1984 of a new process would cause a cgroup policy to be violated.
1990 The "cpuset" controller provides a mechanism for constraining
1991 the CPU and memory node placement of tasks to only the resources
1992 specified in the cpuset interface files in a task's current cgroup.
1993 This is especially valuable on large NUMA systems where placing jobs
1994 on properly sized subsets of the systems with careful processor and
1995 memory placement to reduce cross-node memory access and contention
1996 can improve overall system performance.
1998 The "cpuset" controller is hierarchical. That means the controller
1999 cannot use CPUs or memory nodes not allowed in its parent.
2002 Cpuset Interface Files
2003 ~~~~~~~~~~~~~~~~~~~~~~
2006 A read-write multiple values file which exists on non-root
2007 cpuset-enabled cgroups.
2009 It lists the requested CPUs to be used by tasks within this
2010 cgroup. The actual list of CPUs to be granted, however, is
2011 subjected to constraints imposed by its parent and can differ
2012 from the requested CPUs.
2014 The CPU numbers are comma-separated numbers or ranges.
2020 An empty value indicates that the cgroup is using the same
2021 setting as the nearest cgroup ancestor with a non-empty
2022 "cpuset.cpus" or all the available CPUs if none is found.
2024 The value of "cpuset.cpus" stays constant until the next update
2025 and won't be affected by any CPU hotplug events.
2027 cpuset.cpus.effective
2028 A read-only multiple values file which exists on all
2029 cpuset-enabled cgroups.
2031 It lists the onlined CPUs that are actually granted to this
2032 cgroup by its parent. These CPUs are allowed to be used by
2033 tasks within the current cgroup.
2035 If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows
2036 all the CPUs from the parent cgroup that can be available to
2037 be used by this cgroup. Otherwise, it should be a subset of
2038 "cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus"
2039 can be granted. In this case, it will be treated just like an
2040 empty "cpuset.cpus".
2042 Its value will be affected by CPU hotplug events.
2045 A read-write multiple values file which exists on non-root
2046 cpuset-enabled cgroups.
2048 It lists the requested memory nodes to be used by tasks within
2049 this cgroup. The actual list of memory nodes granted, however,
2050 is subjected to constraints imposed by its parent and can differ
2051 from the requested memory nodes.
2053 The memory node numbers are comma-separated numbers or ranges.
2059 An empty value indicates that the cgroup is using the same
2060 setting as the nearest cgroup ancestor with a non-empty
2061 "cpuset.mems" or all the available memory nodes if none
2064 The value of "cpuset.mems" stays constant until the next update
2065 and won't be affected by any memory nodes hotplug events.
2067 Setting a non-empty value to "cpuset.mems" causes memory of
2068 tasks within the cgroup to be migrated to the designated nodes if
2069 they are currently using memory outside of the designated nodes.
2071 There is a cost for this memory migration. The migration
2072 may not be complete and some memory pages may be left behind.
2073 So it is recommended that "cpuset.mems" should be set properly
2074 before spawning new tasks into the cpuset. Even if there is
2075 a need to change "cpuset.mems" with active tasks, it shouldn't
2078 cpuset.mems.effective
2079 A read-only multiple values file which exists on all
2080 cpuset-enabled cgroups.
2082 It lists the onlined memory nodes that are actually granted to
2083 this cgroup by its parent. These memory nodes are allowed to
2084 be used by tasks within the current cgroup.
2086 If "cpuset.mems" is empty, it shows all the memory nodes from the
2087 parent cgroup that will be available to be used by this cgroup.
2088 Otherwise, it should be a subset of "cpuset.mems" unless none of
2089 the memory nodes listed in "cpuset.mems" can be granted. In this
2090 case, it will be treated just like an empty "cpuset.mems".
2092 Its value will be affected by memory nodes hotplug events.
2094 cpuset.cpus.partition
2095 A read-write single value file which exists on non-root
2096 cpuset-enabled cgroups. This flag is owned by the parent cgroup
2097 and is not delegatable.
2099 It accepts only the following input values when written to.
2101 ======== ================================
2102 "root" a partition root
2103 "member" a non-root member of a partition
2104 ======== ================================
2106 When set to be a partition root, the current cgroup is the
2107 root of a new partition or scheduling domain that comprises
2108 itself and all its descendants except those that are separate
2109 partition roots themselves and their descendants. The root
2110 cgroup is always a partition root.
2112 There are constraints on where a partition root can be set.
2113 It can only be set in a cgroup if all the following conditions
2116 1) The "cpuset.cpus" is not empty and the list of CPUs are
2117 exclusive, i.e. they are not shared by any of its siblings.
2118 2) The parent cgroup is a partition root.
2119 3) The "cpuset.cpus" is also a proper subset of the parent's
2120 "cpuset.cpus.effective".
2121 4) There is no child cgroups with cpuset enabled. This is for
2122 eliminating corner cases that have to be handled if such a
2123 condition is allowed.
2125 Setting it to partition root will take the CPUs away from the
2126 effective CPUs of the parent cgroup. Once it is set, this
2127 file cannot be reverted back to "member" if there are any child
2128 cgroups with cpuset enabled.
2130 A parent partition cannot distribute all its CPUs to its
2131 child partitions. There must be at least one cpu left in the
2134 Once becoming a partition root, changes to "cpuset.cpus" is
2135 generally allowed as long as the first condition above is true,
2136 the change will not take away all the CPUs from the parent
2137 partition and the new "cpuset.cpus" value is a superset of its
2138 children's "cpuset.cpus" values.
2140 Sometimes, external factors like changes to ancestors'
2141 "cpuset.cpus" or cpu hotplug can cause the state of the partition
2142 root to change. On read, the "cpuset.sched.partition" file
2143 can show the following values.
2145 ============== ==============================
2146 "member" Non-root member of a partition
2147 "root" Partition root
2148 "root invalid" Invalid partition root
2149 ============== ==============================
2151 It is a partition root if the first 2 partition root conditions
2152 above are true and at least one CPU from "cpuset.cpus" is
2153 granted by the parent cgroup.
2155 A partition root can become invalid if none of CPUs requested
2156 in "cpuset.cpus" can be granted by the parent cgroup or the
2157 parent cgroup is no longer a partition root itself. In this
2158 case, it is not a real partition even though the restriction
2159 of the first partition root condition above will still apply.
2160 The cpu affinity of all the tasks in the cgroup will then be
2161 associated with CPUs in the nearest ancestor partition.
2163 An invalid partition root can be transitioned back to a
2164 real partition root if at least one of the requested CPUs
2165 can now be granted by its parent. In this case, the cpu
2166 affinity of all the tasks in the formerly invalid partition
2167 will be associated to the CPUs of the newly formed partition.
2168 Changing the partition state of an invalid partition root to
2169 "member" is always allowed even if child cpusets are present.
2175 Device controller manages access to device files. It includes both
2176 creation of new device files (using mknod), and access to the
2177 existing device files.
2179 Cgroup v2 device controller has no interface files and is implemented
2180 on top of cgroup BPF. To control access to device files, a user may
2181 create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach
2182 them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a
2183 device file, corresponding BPF programs will be executed, and depending
2184 on the return value the attempt will succeed or fail with -EPERM.
2186 A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the
2187 bpf_cgroup_dev_ctx structure, which describes the device access attempt:
2188 access type (mknod/read/write) and device (type, major and minor numbers).
2189 If the program returns 0, the attempt fails with -EPERM, otherwise it
2192 An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in
2193 tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree.
2199 The "rdma" controller regulates the distribution and accounting of
2202 RDMA Interface Files
2203 ~~~~~~~~~~~~~~~~~~~~
2206 A readwrite nested-keyed file that exists for all the cgroups
2207 except root that describes current configured resource limit
2208 for a RDMA/IB device.
2210 Lines are keyed by device name and are not ordered.
2211 Each line contains space separated resource name and its configured
2212 limit that can be distributed.
2214 The following nested keys are defined.
2216 ========== =============================
2217 hca_handle Maximum number of HCA Handles
2218 hca_object Maximum number of HCA Objects
2219 ========== =============================
2221 An example for mlx4 and ocrdma device follows::
2223 mlx4_0 hca_handle=2 hca_object=2000
2224 ocrdma1 hca_handle=3 hca_object=max
2227 A read-only file that describes current resource usage.
2228 It exists for all the cgroup except root.
2230 An example for mlx4 and ocrdma device follows::
2232 mlx4_0 hca_handle=1 hca_object=20
2233 ocrdma1 hca_handle=1 hca_object=23
2238 The HugeTLB controller allows to limit the HugeTLB usage per control group and
2239 enforces the controller limit during page fault.
2241 HugeTLB Interface Files
2242 ~~~~~~~~~~~~~~~~~~~~~~~
2244 hugetlb.<hugepagesize>.current
2245 Show current usage for "hugepagesize" hugetlb. It exists for all
2246 the cgroup except root.
2248 hugetlb.<hugepagesize>.max
2249 Set/show the hard limit of "hugepagesize" hugetlb usage.
2250 The default value is "max". It exists for all the cgroup except root.
2252 hugetlb.<hugepagesize>.events
2253 A read-only flat-keyed file which exists on non-root cgroups.
2256 The number of allocation failure due to HugeTLB limit
2258 hugetlb.<hugepagesize>.events.local
2259 Similar to hugetlb.<hugepagesize>.events but the fields in the file
2260 are local to the cgroup i.e. not hierarchical. The file modified event
2261 generated on this file reflects only the local events.
2266 The Miscellaneous cgroup provides the resource limiting and tracking
2267 mechanism for the scalar resources which cannot be abstracted like the other
2268 cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config
2271 A resource can be added to the controller via enum misc_res_type{} in the
2272 include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[]
2273 in the kernel/cgroup/misc.c file. Provider of the resource must set its
2274 capacity prior to using the resource by calling misc_cg_set_capacity().
2276 Once a capacity is set then the resource usage can be updated using charge and
2277 uncharge APIs. All of the APIs to interact with misc controller are in
2278 include/linux/misc_cgroup.h.
2280 Misc Interface Files
2281 ~~~~~~~~~~~~~~~~~~~~
2283 Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then:
2286 A read-only flat-keyed file shown only in the root cgroup. It shows
2287 miscellaneous scalar resources available on the platform along with
2295 A read-only flat-keyed file shown in the non-root cgroups. It shows
2296 the current usage of the resources in the cgroup and its children.::
2303 A read-write flat-keyed file shown in the non root cgroups. Allowed
2304 maximum usage of the resources in the cgroup and its children.::
2310 Limit can be set by::
2312 # echo res_a 1 > misc.max
2314 Limit can be set to max by::
2316 # echo res_a max > misc.max
2318 Limits can be set higher than the capacity value in the misc.capacity
2322 A read-only flat-keyed file which exists on non-root cgroups. The
2323 following entries are defined. Unless specified otherwise, a value
2324 change in this file generates a file modified event. All fields in
2325 this file are hierarchical.
2328 The number of times the cgroup's resource usage was
2329 about to go over the max boundary.
2331 Migration and Ownership
2332 ~~~~~~~~~~~~~~~~~~~~~~~
2334 A miscellaneous scalar resource is charged to the cgroup in which it is used
2335 first, and stays charged to that cgroup until that resource is freed. Migrating
2336 a process to a different cgroup does not move the charge to the destination
2337 cgroup where the process has moved.
2345 perf_event controller, if not mounted on a legacy hierarchy, is
2346 automatically enabled on the v2 hierarchy so that perf events can
2347 always be filtered by cgroup v2 path. The controller can still be
2348 moved to a legacy hierarchy after v2 hierarchy is populated.
2351 Non-normative information
2352 -------------------------
2354 This section contains information that isn't considered to be a part of
2355 the stable kernel API and so is subject to change.
2358 CPU controller root cgroup process behaviour
2359 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2361 When distributing CPU cycles in the root cgroup each thread in this
2362 cgroup is treated as if it was hosted in a separate child cgroup of the
2363 root cgroup. This child cgroup weight is dependent on its thread nice
2366 For details of this mapping see sched_prio_to_weight array in
2367 kernel/sched/core.c file (values from this array should be scaled
2368 appropriately so the neutral - nice 0 - value is 100 instead of 1024).
2371 IO controller root cgroup process behaviour
2372 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2374 Root cgroup processes are hosted in an implicit leaf child node.
2375 When distributing IO resources this implicit child node is taken into
2376 account as if it was a normal child cgroup of the root cgroup with a
2377 weight value of 200.
2386 cgroup namespace provides a mechanism to virtualize the view of the
2387 "/proc/$PID/cgroup" file and cgroup mounts. The CLONE_NEWCGROUP clone
2388 flag can be used with clone(2) and unshare(2) to create a new cgroup
2389 namespace. The process running inside the cgroup namespace will have
2390 its "/proc/$PID/cgroup" output restricted to cgroupns root. The
2391 cgroupns root is the cgroup of the process at the time of creation of
2392 the cgroup namespace.
2394 Without cgroup namespace, the "/proc/$PID/cgroup" file shows the
2395 complete path of the cgroup of a process. In a container setup where
2396 a set of cgroups and namespaces are intended to isolate processes the
2397 "/proc/$PID/cgroup" file may leak potential system level information
2398 to the isolated processes. For example::
2400 # cat /proc/self/cgroup
2401 0::/batchjobs/container_id1
2403 The path '/batchjobs/container_id1' can be considered as system-data
2404 and undesirable to expose to the isolated processes. cgroup namespace
2405 can be used to restrict visibility of this path. For example, before
2406 creating a cgroup namespace, one would see::
2408 # ls -l /proc/self/ns/cgroup
2409 lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835]
2410 # cat /proc/self/cgroup
2411 0::/batchjobs/container_id1
2413 After unsharing a new namespace, the view changes::
2415 # ls -l /proc/self/ns/cgroup
2416 lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183]
2417 # cat /proc/self/cgroup
2420 When some thread from a multi-threaded process unshares its cgroup
2421 namespace, the new cgroupns gets applied to the entire process (all
2422 the threads). This is natural for the v2 hierarchy; however, for the
2423 legacy hierarchies, this may be unexpected.
2425 A cgroup namespace is alive as long as there are processes inside or
2426 mounts pinning it. When the last usage goes away, the cgroup
2427 namespace is destroyed. The cgroupns root and the actual cgroups
2434 The 'cgroupns root' for a cgroup namespace is the cgroup in which the
2435 process calling unshare(2) is running. For example, if a process in
2436 /batchjobs/container_id1 cgroup calls unshare, cgroup
2437 /batchjobs/container_id1 becomes the cgroupns root. For the
2438 init_cgroup_ns, this is the real root ('/') cgroup.
2440 The cgroupns root cgroup does not change even if the namespace creator
2441 process later moves to a different cgroup::
2443 # ~/unshare -c # unshare cgroupns in some cgroup
2444 # cat /proc/self/cgroup
2447 # echo 0 > sub_cgrp_1/cgroup.procs
2448 # cat /proc/self/cgroup
2451 Each process gets its namespace-specific view of "/proc/$PID/cgroup"
2453 Processes running inside the cgroup namespace will be able to see
2454 cgroup paths (in /proc/self/cgroup) only inside their root cgroup.
2455 From within an unshared cgroupns::
2459 # echo 7353 > sub_cgrp_1/cgroup.procs
2460 # cat /proc/7353/cgroup
2463 From the initial cgroup namespace, the real cgroup path will be
2466 $ cat /proc/7353/cgroup
2467 0::/batchjobs/container_id1/sub_cgrp_1
2469 From a sibling cgroup namespace (that is, a namespace rooted at a
2470 different cgroup), the cgroup path relative to its own cgroup
2471 namespace root will be shown. For instance, if PID 7353's cgroup
2472 namespace root is at '/batchjobs/container_id2', then it will see::
2474 # cat /proc/7353/cgroup
2475 0::/../container_id2/sub_cgrp_1
2477 Note that the relative path always starts with '/' to indicate that
2478 its relative to the cgroup namespace root of the caller.
2481 Migration and setns(2)
2482 ----------------------
2484 Processes inside a cgroup namespace can move into and out of the
2485 namespace root if they have proper access to external cgroups. For
2486 example, from inside a namespace with cgroupns root at
2487 /batchjobs/container_id1, and assuming that the global hierarchy is
2488 still accessible inside cgroupns::
2490 # cat /proc/7353/cgroup
2492 # echo 7353 > batchjobs/container_id2/cgroup.procs
2493 # cat /proc/7353/cgroup
2494 0::/../container_id2
2496 Note that this kind of setup is not encouraged. A task inside cgroup
2497 namespace should only be exposed to its own cgroupns hierarchy.
2499 setns(2) to another cgroup namespace is allowed when:
2501 (a) the process has CAP_SYS_ADMIN against its current user namespace
2502 (b) the process has CAP_SYS_ADMIN against the target cgroup
2505 No implicit cgroup changes happen with attaching to another cgroup
2506 namespace. It is expected that the someone moves the attaching
2507 process under the target cgroup namespace root.
2510 Interaction with Other Namespaces
2511 ---------------------------------
2513 Namespace specific cgroup hierarchy can be mounted by a process
2514 running inside a non-init cgroup namespace::
2516 # mount -t cgroup2 none $MOUNT_POINT
2518 This will mount the unified cgroup hierarchy with cgroupns root as the
2519 filesystem root. The process needs CAP_SYS_ADMIN against its user and
2522 The virtualization of /proc/self/cgroup file combined with restricting
2523 the view of cgroup hierarchy by namespace-private cgroupfs mount
2524 provides a properly isolated cgroup view inside the container.
2527 Information on Kernel Programming
2528 =================================
2530 This section contains kernel programming information in the areas
2531 where interacting with cgroup is necessary. cgroup core and
2532 controllers are not covered.
2535 Filesystem Support for Writeback
2536 --------------------------------
2538 A filesystem can support cgroup writeback by updating
2539 address_space_operations->writepage[s]() to annotate bio's using the
2540 following two functions.
2542 wbc_init_bio(@wbc, @bio)
2543 Should be called for each bio carrying writeback data and
2544 associates the bio with the inode's owner cgroup and the
2545 corresponding request queue. This must be called after
2546 a queue (device) has been associated with the bio and
2549 wbc_account_cgroup_owner(@wbc, @page, @bytes)
2550 Should be called for each data segment being written out.
2551 While this function doesn't care exactly when it's called
2552 during the writeback session, it's the easiest and most
2553 natural to call it as data segments are added to a bio.
2555 With writeback bio's annotated, cgroup support can be enabled per
2556 super_block by setting SB_I_CGROUPWB in ->s_iflags. This allows for
2557 selective disabling of cgroup writeback support which is helpful when
2558 certain filesystem features, e.g. journaled data mode, are
2561 wbc_init_bio() binds the specified bio to its cgroup. Depending on
2562 the configuration, the bio may be executed at a lower priority and if
2563 the writeback session is holding shared resources, e.g. a journal
2564 entry, may lead to priority inversion. There is no one easy solution
2565 for the problem. Filesystems can try to work around specific problem
2566 cases by skipping wbc_init_bio() and using bio_associate_blkg()
2570 Deprecated v1 Core Features
2571 ===========================
2573 - Multiple hierarchies including named ones are not supported.
2575 - All v1 mount options are not supported.
2577 - The "tasks" file is removed and "cgroup.procs" is not sorted.
2579 - "cgroup.clone_children" is removed.
2581 - /proc/cgroups is meaningless for v2. Use "cgroup.controllers" file
2582 at the root instead.
2585 Issues with v1 and Rationales for v2
2586 ====================================
2588 Multiple Hierarchies
2589 --------------------
2591 cgroup v1 allowed an arbitrary number of hierarchies and each
2592 hierarchy could host any number of controllers. While this seemed to
2593 provide a high level of flexibility, it wasn't useful in practice.
2595 For example, as there is only one instance of each controller, utility
2596 type controllers such as freezer which can be useful in all
2597 hierarchies could only be used in one. The issue is exacerbated by
2598 the fact that controllers couldn't be moved to another hierarchy once
2599 hierarchies were populated. Another issue was that all controllers
2600 bound to a hierarchy were forced to have exactly the same view of the
2601 hierarchy. It wasn't possible to vary the granularity depending on
2602 the specific controller.
2604 In practice, these issues heavily limited which controllers could be
2605 put on the same hierarchy and most configurations resorted to putting
2606 each controller on its own hierarchy. Only closely related ones, such
2607 as the cpu and cpuacct controllers, made sense to be put on the same
2608 hierarchy. This often meant that userland ended up managing multiple
2609 similar hierarchies repeating the same steps on each hierarchy
2610 whenever a hierarchy management operation was necessary.
2612 Furthermore, support for multiple hierarchies came at a steep cost.
2613 It greatly complicated cgroup core implementation but more importantly
2614 the support for multiple hierarchies restricted how cgroup could be
2615 used in general and what controllers was able to do.
2617 There was no limit on how many hierarchies there might be, which meant
2618 that a thread's cgroup membership couldn't be described in finite
2619 length. The key might contain any number of entries and was unlimited
2620 in length, which made it highly awkward to manipulate and led to
2621 addition of controllers which existed only to identify membership,
2622 which in turn exacerbated the original problem of proliferating number
2625 Also, as a controller couldn't have any expectation regarding the
2626 topologies of hierarchies other controllers might be on, each
2627 controller had to assume that all other controllers were attached to
2628 completely orthogonal hierarchies. This made it impossible, or at
2629 least very cumbersome, for controllers to cooperate with each other.
2631 In most use cases, putting controllers on hierarchies which are
2632 completely orthogonal to each other isn't necessary. What usually is
2633 called for is the ability to have differing levels of granularity
2634 depending on the specific controller. In other words, hierarchy may
2635 be collapsed from leaf towards root when viewed from specific
2636 controllers. For example, a given configuration might not care about
2637 how memory is distributed beyond a certain level while still wanting
2638 to control how CPU cycles are distributed.
2644 cgroup v1 allowed threads of a process to belong to different cgroups.
2645 This didn't make sense for some controllers and those controllers
2646 ended up implementing different ways to ignore such situations but
2647 much more importantly it blurred the line between API exposed to
2648 individual applications and system management interface.
2650 Generally, in-process knowledge is available only to the process
2651 itself; thus, unlike service-level organization of processes,
2652 categorizing threads of a process requires active participation from
2653 the application which owns the target process.
2655 cgroup v1 had an ambiguously defined delegation model which got abused
2656 in combination with thread granularity. cgroups were delegated to
2657 individual applications so that they can create and manage their own
2658 sub-hierarchies and control resource distributions along them. This
2659 effectively raised cgroup to the status of a syscall-like API exposed
2662 First of all, cgroup has a fundamentally inadequate interface to be
2663 exposed this way. For a process to access its own knobs, it has to
2664 extract the path on the target hierarchy from /proc/self/cgroup,
2665 construct the path by appending the name of the knob to the path, open
2666 and then read and/or write to it. This is not only extremely clunky
2667 and unusual but also inherently racy. There is no conventional way to
2668 define transaction across the required steps and nothing can guarantee
2669 that the process would actually be operating on its own sub-hierarchy.
2671 cgroup controllers implemented a number of knobs which would never be
2672 accepted as public APIs because they were just adding control knobs to
2673 system-management pseudo filesystem. cgroup ended up with interface
2674 knobs which were not properly abstracted or refined and directly
2675 revealed kernel internal details. These knobs got exposed to
2676 individual applications through the ill-defined delegation mechanism
2677 effectively abusing cgroup as a shortcut to implementing public APIs
2678 without going through the required scrutiny.
2680 This was painful for both userland and kernel. Userland ended up with
2681 misbehaving and poorly abstracted interfaces and kernel exposing and
2682 locked into constructs inadvertently.
2685 Competition Between Inner Nodes and Threads
2686 -------------------------------------------
2688 cgroup v1 allowed threads to be in any cgroups which created an
2689 interesting problem where threads belonging to a parent cgroup and its
2690 children cgroups competed for resources. This was nasty as two
2691 different types of entities competed and there was no obvious way to
2692 settle it. Different controllers did different things.
2694 The cpu controller considered threads and cgroups as equivalents and
2695 mapped nice levels to cgroup weights. This worked for some cases but
2696 fell flat when children wanted to be allocated specific ratios of CPU
2697 cycles and the number of internal threads fluctuated - the ratios
2698 constantly changed as the number of competing entities fluctuated.
2699 There also were other issues. The mapping from nice level to weight
2700 wasn't obvious or universal, and there were various other knobs which
2701 simply weren't available for threads.
2703 The io controller implicitly created a hidden leaf node for each
2704 cgroup to host the threads. The hidden leaf had its own copies of all
2705 the knobs with ``leaf_`` prefixed. While this allowed equivalent
2706 control over internal threads, it was with serious drawbacks. It
2707 always added an extra layer of nesting which wouldn't be necessary
2708 otherwise, made the interface messy and significantly complicated the
2711 The memory controller didn't have a way to control what happened
2712 between internal tasks and child cgroups and the behavior was not
2713 clearly defined. There were attempts to add ad-hoc behaviors and
2714 knobs to tailor the behavior to specific workloads which would have
2715 led to problems extremely difficult to resolve in the long term.
2717 Multiple controllers struggled with internal tasks and came up with
2718 different ways to deal with it; unfortunately, all the approaches were
2719 severely flawed and, furthermore, the widely different behaviors
2720 made cgroup as a whole highly inconsistent.
2722 This clearly is a problem which needs to be addressed from cgroup core
2726 Other Interface Issues
2727 ----------------------
2729 cgroup v1 grew without oversight and developed a large number of
2730 idiosyncrasies and inconsistencies. One issue on the cgroup core side
2731 was how an empty cgroup was notified - a userland helper binary was
2732 forked and executed for each event. The event delivery wasn't
2733 recursive or delegatable. The limitations of the mechanism also led
2734 to in-kernel event delivery filtering mechanism further complicating
2737 Controller interfaces were problematic too. An extreme example is
2738 controllers completely ignoring hierarchical organization and treating
2739 all cgroups as if they were all located directly under the root
2740 cgroup. Some controllers exposed a large amount of inconsistent
2741 implementation details to userland.
2743 There also was no consistency across controllers. When a new cgroup
2744 was created, some controllers defaulted to not imposing extra
2745 restrictions while others disallowed any resource usage until
2746 explicitly configured. Configuration knobs for the same type of
2747 control used widely differing naming schemes and formats. Statistics
2748 and information knobs were named arbitrarily and used different
2749 formats and units even in the same controller.
2751 cgroup v2 establishes common conventions where appropriate and updates
2752 controllers so that they expose minimal and consistent interfaces.
2755 Controller Issues and Remedies
2756 ------------------------------
2761 The original lower boundary, the soft limit, is defined as a limit
2762 that is per default unset. As a result, the set of cgroups that
2763 global reclaim prefers is opt-in, rather than opt-out. The costs for
2764 optimizing these mostly negative lookups are so high that the
2765 implementation, despite its enormous size, does not even provide the
2766 basic desirable behavior. First off, the soft limit has no
2767 hierarchical meaning. All configured groups are organized in a global
2768 rbtree and treated like equal peers, regardless where they are located
2769 in the hierarchy. This makes subtree delegation impossible. Second,
2770 the soft limit reclaim pass is so aggressive that it not just
2771 introduces high allocation latencies into the system, but also impacts
2772 system performance due to overreclaim, to the point where the feature
2773 becomes self-defeating.
2775 The memory.low boundary on the other hand is a top-down allocated
2776 reserve. A cgroup enjoys reclaim protection when it's within its
2777 effective low, which makes delegation of subtrees possible. It also
2778 enjoys having reclaim pressure proportional to its overage when
2779 above its effective low.
2781 The original high boundary, the hard limit, is defined as a strict
2782 limit that can not budge, even if the OOM killer has to be called.
2783 But this generally goes against the goal of making the most out of the
2784 available memory. The memory consumption of workloads varies during
2785 runtime, and that requires users to overcommit. But doing that with a
2786 strict upper limit requires either a fairly accurate prediction of the
2787 working set size or adding slack to the limit. Since working set size
2788 estimation is hard and error prone, and getting it wrong results in
2789 OOM kills, most users tend to err on the side of a looser limit and
2790 end up wasting precious resources.
2792 The memory.high boundary on the other hand can be set much more
2793 conservatively. When hit, it throttles allocations by forcing them
2794 into direct reclaim to work off the excess, but it never invokes the
2795 OOM killer. As a result, a high boundary that is chosen too
2796 aggressively will not terminate the processes, but instead it will
2797 lead to gradual performance degradation. The user can monitor this
2798 and make corrections until the minimal memory footprint that still
2799 gives acceptable performance is found.
2801 In extreme cases, with many concurrent allocations and a complete
2802 breakdown of reclaim progress within the group, the high boundary can
2803 be exceeded. But even then it's mostly better to satisfy the
2804 allocation from the slack available in other groups or the rest of the
2805 system than killing the group. Otherwise, memory.max is there to
2806 limit this type of spillover and ultimately contain buggy or even
2807 malicious applications.
2809 Setting the original memory.limit_in_bytes below the current usage was
2810 subject to a race condition, where concurrent charges could cause the
2811 limit setting to fail. memory.max on the other hand will first set the
2812 limit to prevent new charges, and then reclaim and OOM kill until the
2813 new limit is met - or the task writing to memory.max is killed.
2815 The combined memory+swap accounting and limiting is replaced by real
2816 control over swap space.
2818 The main argument for a combined memory+swap facility in the original
2819 cgroup design was that global or parental pressure would always be
2820 able to swap all anonymous memory of a child group, regardless of the
2821 child's own (possibly untrusted) configuration. However, untrusted
2822 groups can sabotage swapping by other means - such as referencing its
2823 anonymous memory in a tight loop - and an admin can not assume full
2824 swappability when overcommitting untrusted jobs.
2826 For trusted jobs, on the other hand, a combined counter is not an
2827 intuitive userspace interface, and it flies in the face of the idea
2828 that cgroup controllers should account and limit specific physical
2829 resources. Swap space is a resource like all others in the system,
2830 and that's why unified hierarchy allows distributing it separately.