6 :Author: Tejun Heo <tj@kernel.org>
8 This is the authoritative documentation on the design, interface and
9 conventions of cgroup v2. It describes all userland-visible aspects
10 of cgroup including core and specific controller behaviors. All
11 future changes must be reflected in this document. Documentation for
12 v1 is available under Documentation/admin-guide/cgroup-v1/.
21 2-2. Organizing Processes and Threads
24 2-3. [Un]populated Notification
25 2-4. Controlling Controllers
26 2-4-1. Enabling and Disabling
27 2-4-2. Top-down Constraint
28 2-4-3. No Internal Process Constraint
30 2-5-1. Model of Delegation
31 2-5-2. Delegation Containment
33 2-6-1. Organize Once and Control
34 2-6-2. Avoid Name Collisions
35 3. Resource Distribution Models
43 4-3. Core Interface Files
46 5-1-1. CPU Interface Files
48 5-2-1. Memory Interface Files
49 5-2-2. Usage Guidelines
50 5-2-3. Memory Ownership
52 5-3-1. IO Interface Files
55 5-3-3-1. How IO Latency Throttling Works
56 5-3-3-2. IO Latency Interface Files
58 5-4-1. PID Interface Files
60 5.5-1. Cpuset Interface Files
63 5-7-1. RDMA Interface Files
66 5-N. Non-normative information
67 5-N-1. CPU controller root cgroup process behaviour
68 5-N-2. IO controller root cgroup process behaviour
71 6-2. The Root and Views
72 6-3. Migration and setns(2)
73 6-4. Interaction with Other Namespaces
74 P. Information on Kernel Programming
75 P-1. Filesystem Support for Writeback
76 D. Deprecated v1 Core Features
77 R. Issues with v1 and Rationales for v2
78 R-1. Multiple Hierarchies
79 R-2. Thread Granularity
80 R-3. Competition Between Inner Nodes and Threads
81 R-4. Other Interface Issues
82 R-5. Controller Issues and Remedies
92 "cgroup" stands for "control group" and is never capitalized. The
93 singular form is used to designate the whole feature and also as a
94 qualifier as in "cgroup controllers". When explicitly referring to
95 multiple individual control groups, the plural form "cgroups" is used.
101 cgroup is a mechanism to organize processes hierarchically and
102 distribute system resources along the hierarchy in a controlled and
105 cgroup is largely composed of two parts - the core and controllers.
106 cgroup core is primarily responsible for hierarchically organizing
107 processes. A cgroup controller is usually responsible for
108 distributing a specific type of system resource along the hierarchy
109 although there are utility controllers which serve purposes other than
110 resource distribution.
112 cgroups form a tree structure and every process in the system belongs
113 to one and only one cgroup. All threads of a process belong to the
114 same cgroup. On creation, all processes are put in the cgroup that
115 the parent process belongs to at the time. A process can be migrated
116 to another cgroup. Migration of a process doesn't affect already
117 existing descendant processes.
119 Following certain structural constraints, controllers may be enabled or
120 disabled selectively on a cgroup. All controller behaviors are
121 hierarchical - if a controller is enabled on a cgroup, it affects all
122 processes which belong to the cgroups consisting the inclusive
123 sub-hierarchy of the cgroup. When a controller is enabled on a nested
124 cgroup, it always restricts the resource distribution further. The
125 restrictions set closer to the root in the hierarchy can not be
126 overridden from further away.
135 Unlike v1, cgroup v2 has only single hierarchy. The cgroup v2
136 hierarchy can be mounted with the following mount command::
138 # mount -t cgroup2 none $MOUNT_POINT
140 cgroup2 filesystem has the magic number 0x63677270 ("cgrp"). All
141 controllers which support v2 and are not bound to a v1 hierarchy are
142 automatically bound to the v2 hierarchy and show up at the root.
143 Controllers which are not in active use in the v2 hierarchy can be
144 bound to other hierarchies. This allows mixing v2 hierarchy with the
145 legacy v1 multiple hierarchies in a fully backward compatible way.
147 A controller can be moved across hierarchies only after the controller
148 is no longer referenced in its current hierarchy. Because per-cgroup
149 controller states are destroyed asynchronously and controllers may
150 have lingering references, a controller may not show up immediately on
151 the v2 hierarchy after the final umount of the previous hierarchy.
152 Similarly, a controller should be fully disabled to be moved out of
153 the unified hierarchy and it may take some time for the disabled
154 controller to become available for other hierarchies; furthermore, due
155 to inter-controller dependencies, other controllers may need to be
158 While useful for development and manual configurations, moving
159 controllers dynamically between the v2 and other hierarchies is
160 strongly discouraged for production use. It is recommended to decide
161 the hierarchies and controller associations before starting using the
162 controllers after system boot.
164 During transition to v2, system management software might still
165 automount the v1 cgroup filesystem and so hijack all controllers
166 during boot, before manual intervention is possible. To make testing
167 and experimenting easier, the kernel parameter cgroup_no_v1= allows
168 disabling controllers in v1 and make them always available in v2.
170 cgroup v2 currently supports the following mount options.
174 Consider cgroup namespaces as delegation boundaries. This
175 option is system wide and can only be set on mount or modified
176 through remount from the init namespace. The mount option is
177 ignored on non-init namespace mounts. Please refer to the
178 Delegation section for details.
182 Only populate memory.events with data for the current cgroup,
183 and not any subtrees. This is legacy behaviour, the default
184 behaviour without this option is to include subtree counts.
185 This option is system wide and can only be set on mount or
186 modified through remount from the init namespace. The mount
187 option is ignored on non-init namespace mounts.
190 Organizing Processes and Threads
191 --------------------------------
196 Initially, only the root cgroup exists to which all processes belong.
197 A child cgroup can be created by creating a sub-directory::
201 A given cgroup may have multiple child cgroups forming a tree
202 structure. Each cgroup has a read-writable interface file
203 "cgroup.procs". When read, it lists the PIDs of all processes which
204 belong to the cgroup one-per-line. The PIDs are not ordered and the
205 same PID may show up more than once if the process got moved to
206 another cgroup and then back or the PID got recycled while reading.
208 A process can be migrated into a cgroup by writing its PID to the
209 target cgroup's "cgroup.procs" file. Only one process can be migrated
210 on a single write(2) call. If a process is composed of multiple
211 threads, writing the PID of any thread migrates all threads of the
214 When a process forks a child process, the new process is born into the
215 cgroup that the forking process belongs to at the time of the
216 operation. After exit, a process stays associated with the cgroup
217 that it belonged to at the time of exit until it's reaped; however, a
218 zombie process does not appear in "cgroup.procs" and thus can't be
219 moved to another cgroup.
221 A cgroup which doesn't have any children or live processes can be
222 destroyed by removing the directory. Note that a cgroup which doesn't
223 have any children and is associated only with zombie processes is
224 considered empty and can be removed::
228 "/proc/$PID/cgroup" lists a process's cgroup membership. If legacy
229 cgroup is in use in the system, this file may contain multiple lines,
230 one for each hierarchy. The entry for cgroup v2 is always in the
233 # cat /proc/842/cgroup
235 0::/test-cgroup/test-cgroup-nested
237 If the process becomes a zombie and the cgroup it was associated with
238 is removed subsequently, " (deleted)" is appended to the path::
240 # cat /proc/842/cgroup
242 0::/test-cgroup/test-cgroup-nested (deleted)
248 cgroup v2 supports thread granularity for a subset of controllers to
249 support use cases requiring hierarchical resource distribution across
250 the threads of a group of processes. By default, all threads of a
251 process belong to the same cgroup, which also serves as the resource
252 domain to host resource consumptions which are not specific to a
253 process or thread. The thread mode allows threads to be spread across
254 a subtree while still maintaining the common resource domain for them.
256 Controllers which support thread mode are called threaded controllers.
257 The ones which don't are called domain controllers.
259 Marking a cgroup threaded makes it join the resource domain of its
260 parent as a threaded cgroup. The parent may be another threaded
261 cgroup whose resource domain is further up in the hierarchy. The root
262 of a threaded subtree, that is, the nearest ancestor which is not
263 threaded, is called threaded domain or thread root interchangeably and
264 serves as the resource domain for the entire subtree.
266 Inside a threaded subtree, threads of a process can be put in
267 different cgroups and are not subject to the no internal process
268 constraint - threaded controllers can be enabled on non-leaf cgroups
269 whether they have threads in them or not.
271 As the threaded domain cgroup hosts all the domain resource
272 consumptions of the subtree, it is considered to have internal
273 resource consumptions whether there are processes in it or not and
274 can't have populated child cgroups which aren't threaded. Because the
275 root cgroup is not subject to no internal process constraint, it can
276 serve both as a threaded domain and a parent to domain cgroups.
278 The current operation mode or type of the cgroup is shown in the
279 "cgroup.type" file which indicates whether the cgroup is a normal
280 domain, a domain which is serving as the domain of a threaded subtree,
281 or a threaded cgroup.
283 On creation, a cgroup is always a domain cgroup and can be made
284 threaded by writing "threaded" to the "cgroup.type" file. The
285 operation is single direction::
287 # echo threaded > cgroup.type
289 Once threaded, the cgroup can't be made a domain again. To enable the
290 thread mode, the following conditions must be met.
292 - As the cgroup will join the parent's resource domain. The parent
293 must either be a valid (threaded) domain or a threaded cgroup.
295 - When the parent is an unthreaded domain, it must not have any domain
296 controllers enabled or populated domain children. The root is
297 exempt from this requirement.
299 Topology-wise, a cgroup can be in an invalid state. Please consider
300 the following topology::
302 A (threaded domain) - B (threaded) - C (domain, just created)
304 C is created as a domain but isn't connected to a parent which can
305 host child domains. C can't be used until it is turned into a
306 threaded cgroup. "cgroup.type" file will report "domain (invalid)" in
307 these cases. Operations which fail due to invalid topology use
308 EOPNOTSUPP as the errno.
310 A domain cgroup is turned into a threaded domain when one of its child
311 cgroup becomes threaded or threaded controllers are enabled in the
312 "cgroup.subtree_control" file while there are processes in the cgroup.
313 A threaded domain reverts to a normal domain when the conditions
316 When read, "cgroup.threads" contains the list of the thread IDs of all
317 threads in the cgroup. Except that the operations are per-thread
318 instead of per-process, "cgroup.threads" has the same format and
319 behaves the same way as "cgroup.procs". While "cgroup.threads" can be
320 written to in any cgroup, as it can only move threads inside the same
321 threaded domain, its operations are confined inside each threaded
324 The threaded domain cgroup serves as the resource domain for the whole
325 subtree, and, while the threads can be scattered across the subtree,
326 all the processes are considered to be in the threaded domain cgroup.
327 "cgroup.procs" in a threaded domain cgroup contains the PIDs of all
328 processes in the subtree and is not readable in the subtree proper.
329 However, "cgroup.procs" can be written to from anywhere in the subtree
330 to migrate all threads of the matching process to the cgroup.
332 Only threaded controllers can be enabled in a threaded subtree. When
333 a threaded controller is enabled inside a threaded subtree, it only
334 accounts for and controls resource consumptions associated with the
335 threads in the cgroup and its descendants. All consumptions which
336 aren't tied to a specific thread belong to the threaded domain cgroup.
338 Because a threaded subtree is exempt from no internal process
339 constraint, a threaded controller must be able to handle competition
340 between threads in a non-leaf cgroup and its child cgroups. Each
341 threaded controller defines how such competitions are handled.
344 [Un]populated Notification
345 --------------------------
347 Each non-root cgroup has a "cgroup.events" file which contains
348 "populated" field indicating whether the cgroup's sub-hierarchy has
349 live processes in it. Its value is 0 if there is no live process in
350 the cgroup and its descendants; otherwise, 1. poll and [id]notify
351 events are triggered when the value changes. This can be used, for
352 example, to start a clean-up operation after all processes of a given
353 sub-hierarchy have exited. The populated state updates and
354 notifications are recursive. Consider the following sub-hierarchy
355 where the numbers in the parentheses represent the numbers of processes
361 A, B and C's "populated" fields would be 1 while D's 0. After the one
362 process in C exits, B and C's "populated" fields would flip to "0" and
363 file modified events will be generated on the "cgroup.events" files of
367 Controlling Controllers
368 -----------------------
370 Enabling and Disabling
371 ~~~~~~~~~~~~~~~~~~~~~~
373 Each cgroup has a "cgroup.controllers" file which lists all
374 controllers available for the cgroup to enable::
376 # cat cgroup.controllers
379 No controller is enabled by default. Controllers can be enabled and
380 disabled by writing to the "cgroup.subtree_control" file::
382 # echo "+cpu +memory -io" > cgroup.subtree_control
384 Only controllers which are listed in "cgroup.controllers" can be
385 enabled. When multiple operations are specified as above, either they
386 all succeed or fail. If multiple operations on the same controller
387 are specified, the last one is effective.
389 Enabling a controller in a cgroup indicates that the distribution of
390 the target resource across its immediate children will be controlled.
391 Consider the following sub-hierarchy. The enabled controllers are
392 listed in parentheses::
394 A(cpu,memory) - B(memory) - C()
397 As A has "cpu" and "memory" enabled, A will control the distribution
398 of CPU cycles and memory to its children, in this case, B. As B has
399 "memory" enabled but not "CPU", C and D will compete freely on CPU
400 cycles but their division of memory available to B will be controlled.
402 As a controller regulates the distribution of the target resource to
403 the cgroup's children, enabling it creates the controller's interface
404 files in the child cgroups. In the above example, enabling "cpu" on B
405 would create the "cpu." prefixed controller interface files in C and
406 D. Likewise, disabling "memory" from B would remove the "memory."
407 prefixed controller interface files from C and D. This means that the
408 controller interface files - anything which doesn't start with
409 "cgroup." are owned by the parent rather than the cgroup itself.
415 Resources are distributed top-down and a cgroup can further distribute
416 a resource only if the resource has been distributed to it from the
417 parent. This means that all non-root "cgroup.subtree_control" files
418 can only contain controllers which are enabled in the parent's
419 "cgroup.subtree_control" file. A controller can be enabled only if
420 the parent has the controller enabled and a controller can't be
421 disabled if one or more children have it enabled.
424 No Internal Process Constraint
425 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
427 Non-root cgroups can distribute domain resources to their children
428 only when they don't have any processes of their own. In other words,
429 only domain cgroups which don't contain any processes can have domain
430 controllers enabled in their "cgroup.subtree_control" files.
432 This guarantees that, when a domain controller is looking at the part
433 of the hierarchy which has it enabled, processes are always only on
434 the leaves. This rules out situations where child cgroups compete
435 against internal processes of the parent.
437 The root cgroup is exempt from this restriction. Root contains
438 processes and anonymous resource consumption which can't be associated
439 with any other cgroups and requires special treatment from most
440 controllers. How resource consumption in the root cgroup is governed
441 is up to each controller (for more information on this topic please
442 refer to the Non-normative information section in the Controllers
445 Note that the restriction doesn't get in the way if there is no
446 enabled controller in the cgroup's "cgroup.subtree_control". This is
447 important as otherwise it wouldn't be possible to create children of a
448 populated cgroup. To control resource distribution of a cgroup, the
449 cgroup must create children and transfer all its processes to the
450 children before enabling controllers in its "cgroup.subtree_control"
460 A cgroup can be delegated in two ways. First, to a less privileged
461 user by granting write access of the directory and its "cgroup.procs",
462 "cgroup.threads" and "cgroup.subtree_control" files to the user.
463 Second, if the "nsdelegate" mount option is set, automatically to a
464 cgroup namespace on namespace creation.
466 Because the resource control interface files in a given directory
467 control the distribution of the parent's resources, the delegatee
468 shouldn't be allowed to write to them. For the first method, this is
469 achieved by not granting access to these files. For the second, the
470 kernel rejects writes to all files other than "cgroup.procs" and
471 "cgroup.subtree_control" on a namespace root from inside the
474 The end results are equivalent for both delegation types. Once
475 delegated, the user can build sub-hierarchy under the directory,
476 organize processes inside it as it sees fit and further distribute the
477 resources it received from the parent. The limits and other settings
478 of all resource controllers are hierarchical and regardless of what
479 happens in the delegated sub-hierarchy, nothing can escape the
480 resource restrictions imposed by the parent.
482 Currently, cgroup doesn't impose any restrictions on the number of
483 cgroups in or nesting depth of a delegated sub-hierarchy; however,
484 this may be limited explicitly in the future.
487 Delegation Containment
488 ~~~~~~~~~~~~~~~~~~~~~~
490 A delegated sub-hierarchy is contained in the sense that processes
491 can't be moved into or out of the sub-hierarchy by the delegatee.
493 For delegations to a less privileged user, this is achieved by
494 requiring the following conditions for a process with a non-root euid
495 to migrate a target process into a cgroup by writing its PID to the
498 - The writer must have write access to the "cgroup.procs" file.
500 - The writer must have write access to the "cgroup.procs" file of the
501 common ancestor of the source and destination cgroups.
503 The above two constraints ensure that while a delegatee may migrate
504 processes around freely in the delegated sub-hierarchy it can't pull
505 in from or push out to outside the sub-hierarchy.
507 For an example, let's assume cgroups C0 and C1 have been delegated to
508 user U0 who created C00, C01 under C0 and C10 under C1 as follows and
509 all processes under C0 and C1 belong to U0::
511 ~~~~~~~~~~~~~ - C0 - C00
514 ~~~~~~~~~~~~~ - C1 - C10
516 Let's also say U0 wants to write the PID of a process which is
517 currently in C10 into "C00/cgroup.procs". U0 has write access to the
518 file; however, the common ancestor of the source cgroup C10 and the
519 destination cgroup C00 is above the points of delegation and U0 would
520 not have write access to its "cgroup.procs" files and thus the write
521 will be denied with -EACCES.
523 For delegations to namespaces, containment is achieved by requiring
524 that both the source and destination cgroups are reachable from the
525 namespace of the process which is attempting the migration. If either
526 is not reachable, the migration is rejected with -ENOENT.
532 Organize Once and Control
533 ~~~~~~~~~~~~~~~~~~~~~~~~~
535 Migrating a process across cgroups is a relatively expensive operation
536 and stateful resources such as memory are not moved together with the
537 process. This is an explicit design decision as there often exist
538 inherent trade-offs between migration and various hot paths in terms
539 of synchronization cost.
541 As such, migrating processes across cgroups frequently as a means to
542 apply different resource restrictions is discouraged. A workload
543 should be assigned to a cgroup according to the system's logical and
544 resource structure once on start-up. Dynamic adjustments to resource
545 distribution can be made by changing controller configuration through
549 Avoid Name Collisions
550 ~~~~~~~~~~~~~~~~~~~~~
552 Interface files for a cgroup and its children cgroups occupy the same
553 directory and it is possible to create children cgroups which collide
554 with interface files.
556 All cgroup core interface files are prefixed with "cgroup." and each
557 controller's interface files are prefixed with the controller name and
558 a dot. A controller's name is composed of lower case alphabets and
559 '_'s but never begins with an '_' so it can be used as the prefix
560 character for collision avoidance. Also, interface file names won't
561 start or end with terms which are often used in categorizing workloads
562 such as job, service, slice, unit or workload.
564 cgroup doesn't do anything to prevent name collisions and it's the
565 user's responsibility to avoid them.
568 Resource Distribution Models
569 ============================
571 cgroup controllers implement several resource distribution schemes
572 depending on the resource type and expected use cases. This section
573 describes major schemes in use along with their expected behaviors.
579 A parent's resource is distributed by adding up the weights of all
580 active children and giving each the fraction matching the ratio of its
581 weight against the sum. As only children which can make use of the
582 resource at the moment participate in the distribution, this is
583 work-conserving. Due to the dynamic nature, this model is usually
584 used for stateless resources.
586 All weights are in the range [1, 10000] with the default at 100. This
587 allows symmetric multiplicative biases in both directions at fine
588 enough granularity while staying in the intuitive range.
590 As long as the weight is in range, all configuration combinations are
591 valid and there is no reason to reject configuration changes or
594 "cpu.weight" proportionally distributes CPU cycles to active children
595 and is an example of this type.
601 A child can only consume upto the configured amount of the resource.
602 Limits can be over-committed - the sum of the limits of children can
603 exceed the amount of resource available to the parent.
605 Limits are in the range [0, max] and defaults to "max", which is noop.
607 As limits can be over-committed, all configuration combinations are
608 valid and there is no reason to reject configuration changes or
611 "io.max" limits the maximum BPS and/or IOPS that a cgroup can consume
612 on an IO device and is an example of this type.
618 A cgroup is protected upto the configured amount of the resource
619 as long as the usages of all its ancestors are under their
620 protected levels. Protections can be hard guarantees or best effort
621 soft boundaries. Protections can also be over-committed in which case
622 only upto the amount available to the parent is protected among
625 Protections are in the range [0, max] and defaults to 0, which is
628 As protections can be over-committed, all configuration combinations
629 are valid and there is no reason to reject configuration changes or
632 "memory.low" implements best-effort memory protection and is an
633 example of this type.
639 A cgroup is exclusively allocated a certain amount of a finite
640 resource. Allocations can't be over-committed - the sum of the
641 allocations of children can not exceed the amount of resource
642 available to the parent.
644 Allocations are in the range [0, max] and defaults to 0, which is no
647 As allocations can't be over-committed, some configuration
648 combinations are invalid and should be rejected. Also, if the
649 resource is mandatory for execution of processes, process migrations
652 "cpu.rt.max" hard-allocates realtime slices and is an example of this
662 All interface files should be in one of the following formats whenever
665 New-line separated values
666 (when only one value can be written at once)
672 Space separated values
673 (when read-only or multiple values can be written at once)
685 KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01...
686 KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11...
689 For a writable file, the format for writing should generally match
690 reading; however, controllers may allow omitting later fields or
691 implement restricted shortcuts for most common use cases.
693 For both flat and nested keyed files, only the values for a single key
694 can be written at a time. For nested keyed files, the sub key pairs
695 may be specified in any order and not all pairs have to be specified.
701 - Settings for a single feature should be contained in a single file.
703 - The root cgroup should be exempt from resource control and thus
704 shouldn't have resource control interface files. Also,
705 informational files on the root cgroup which end up showing global
706 information available elsewhere shouldn't exist.
708 - The default time unit is microseconds. If a different unit is ever
709 used, an explicit unit suffix must be present.
711 - A parts-per quantity should use a percentage decimal with at least
712 two digit fractional part - e.g. 13.40.
714 - If a controller implements weight based resource distribution, its
715 interface file should be named "weight" and have the range [1,
716 10000] with 100 as the default. The values are chosen to allow
717 enough and symmetric bias in both directions while keeping it
718 intuitive (the default is 100%).
720 - If a controller implements an absolute resource guarantee and/or
721 limit, the interface files should be named "min" and "max"
722 respectively. If a controller implements best effort resource
723 guarantee and/or limit, the interface files should be named "low"
724 and "high" respectively.
726 In the above four control files, the special token "max" should be
727 used to represent upward infinity for both reading and writing.
729 - If a setting has a configurable default value and keyed specific
730 overrides, the default entry should be keyed with "default" and
731 appear as the first entry in the file.
733 The default value can be updated by writing either "default $VAL" or
736 When writing to update a specific override, "default" can be used as
737 the value to indicate removal of the override. Override entries
738 with "default" as the value must not appear when read.
740 For example, a setting which is keyed by major:minor device numbers
741 with integer values may look like the following::
743 # cat cgroup-example-interface-file
747 The default value can be updated by::
749 # echo 125 > cgroup-example-interface-file
753 # echo "default 125" > cgroup-example-interface-file
755 An override can be set by::
757 # echo "8:16 170" > cgroup-example-interface-file
761 # echo "8:0 default" > cgroup-example-interface-file
762 # cat cgroup-example-interface-file
766 - For events which are not very high frequency, an interface file
767 "events" should be created which lists event key value pairs.
768 Whenever a notifiable event happens, file modified event should be
769 generated on the file.
775 All cgroup core files are prefixed with "cgroup."
779 A read-write single value file which exists on non-root
782 When read, it indicates the current type of the cgroup, which
783 can be one of the following values.
785 - "domain" : A normal valid domain cgroup.
787 - "domain threaded" : A threaded domain cgroup which is
788 serving as the root of a threaded subtree.
790 - "domain invalid" : A cgroup which is in an invalid state.
791 It can't be populated or have controllers enabled. It may
792 be allowed to become a threaded cgroup.
794 - "threaded" : A threaded cgroup which is a member of a
797 A cgroup can be turned into a threaded cgroup by writing
798 "threaded" to this file.
801 A read-write new-line separated values file which exists on
804 When read, it lists the PIDs of all processes which belong to
805 the cgroup one-per-line. The PIDs are not ordered and the
806 same PID may show up more than once if the process got moved
807 to another cgroup and then back or the PID got recycled while
810 A PID can be written to migrate the process associated with
811 the PID to the cgroup. The writer should match all of the
812 following conditions.
814 - It must have write access to the "cgroup.procs" file.
816 - It must have write access to the "cgroup.procs" file of the
817 common ancestor of the source and destination cgroups.
819 When delegating a sub-hierarchy, write access to this file
820 should be granted along with the containing directory.
822 In a threaded cgroup, reading this file fails with EOPNOTSUPP
823 as all the processes belong to the thread root. Writing is
824 supported and moves every thread of the process to the cgroup.
827 A read-write new-line separated values file which exists on
830 When read, it lists the TIDs of all threads which belong to
831 the cgroup one-per-line. The TIDs are not ordered and the
832 same TID may show up more than once if the thread got moved to
833 another cgroup and then back or the TID got recycled while
836 A TID can be written to migrate the thread associated with the
837 TID to the cgroup. The writer should match all of the
838 following conditions.
840 - It must have write access to the "cgroup.threads" file.
842 - The cgroup that the thread is currently in must be in the
843 same resource domain as the destination cgroup.
845 - It must have write access to the "cgroup.procs" file of the
846 common ancestor of the source and destination cgroups.
848 When delegating a sub-hierarchy, write access to this file
849 should be granted along with the containing directory.
852 A read-only space separated values file which exists on all
855 It shows space separated list of all controllers available to
856 the cgroup. The controllers are not ordered.
858 cgroup.subtree_control
859 A read-write space separated values file which exists on all
860 cgroups. Starts out empty.
862 When read, it shows space separated list of the controllers
863 which are enabled to control resource distribution from the
864 cgroup to its children.
866 Space separated list of controllers prefixed with '+' or '-'
867 can be written to enable or disable controllers. A controller
868 name prefixed with '+' enables the controller and '-'
869 disables. If a controller appears more than once on the list,
870 the last one is effective. When multiple enable and disable
871 operations are specified, either all succeed or all fail.
874 A read-only flat-keyed file which exists on non-root cgroups.
875 The following entries are defined. Unless specified
876 otherwise, a value change in this file generates a file
880 1 if the cgroup or its descendants contains any live
881 processes; otherwise, 0.
883 1 if the cgroup is frozen; otherwise, 0.
885 cgroup.max.descendants
886 A read-write single value files. The default is "max".
888 Maximum allowed number of descent cgroups.
889 If the actual number of descendants is equal or larger,
890 an attempt to create a new cgroup in the hierarchy will fail.
893 A read-write single value files. The default is "max".
895 Maximum allowed descent depth below the current cgroup.
896 If the actual descent depth is equal or larger,
897 an attempt to create a new child cgroup will fail.
900 A read-only flat-keyed file with the following entries:
903 Total number of visible descendant cgroups.
906 Total number of dying descendant cgroups. A cgroup becomes
907 dying after being deleted by a user. The cgroup will remain
908 in dying state for some time undefined time (which can depend
909 on system load) before being completely destroyed.
911 A process can't enter a dying cgroup under any circumstances,
912 a dying cgroup can't revive.
914 A dying cgroup can consume system resources not exceeding
915 limits, which were active at the moment of cgroup deletion.
918 A read-write single value file which exists on non-root cgroups.
919 Allowed values are "0" and "1". The default is "0".
921 Writing "1" to the file causes freezing of the cgroup and all
922 descendant cgroups. This means that all belonging processes will
923 be stopped and will not run until the cgroup will be explicitly
924 unfrozen. Freezing of the cgroup may take some time; when this action
925 is completed, the "frozen" value in the cgroup.events control file
926 will be updated to "1" and the corresponding notification will be
929 A cgroup can be frozen either by its own settings, or by settings
930 of any ancestor cgroups. If any of ancestor cgroups is frozen, the
931 cgroup will remain frozen.
933 Processes in the frozen cgroup can be killed by a fatal signal.
934 They also can enter and leave a frozen cgroup: either by an explicit
935 move by a user, or if freezing of the cgroup races with fork().
936 If a process is moved to a frozen cgroup, it stops. If a process is
937 moved out of a frozen cgroup, it becomes running.
939 Frozen status of a cgroup doesn't affect any cgroup tree operations:
940 it's possible to delete a frozen (and empty) cgroup, as well as
941 create new sub-cgroups.
949 The "cpu" controllers regulates distribution of CPU cycles. This
950 controller implements weight and absolute bandwidth limit models for
951 normal scheduling policy and absolute bandwidth allocation model for
952 realtime scheduling policy.
954 In all the above models, cycles distribution is defined only on a temporal
955 base and it does not account for the frequency at which tasks are executed.
956 The (optional) utilization clamping support allows to hint the schedutil
957 cpufreq governor about the minimum desired frequency which should always be
958 provided by a CPU, as well as the maximum desired frequency, which should not
959 be exceeded by a CPU.
961 WARNING: cgroup2 doesn't yet support control of realtime processes and
962 the cpu controller can only be enabled when all RT processes are in
963 the root cgroup. Be aware that system management software may already
964 have placed RT processes into nonroot cgroups during the system boot
965 process, and these processes may need to be moved to the root cgroup
966 before the cpu controller can be enabled.
972 All time durations are in microseconds.
975 A read-only flat-keyed file which exists on non-root cgroups.
976 This file exists whether the controller is enabled or not.
978 It always reports the following three stats:
984 and the following three when the controller is enabled:
991 A read-write single value file which exists on non-root
992 cgroups. The default is "100".
994 The weight in the range [1, 10000].
997 A read-write single value file which exists on non-root
998 cgroups. The default is "0".
1000 The nice value is in the range [-20, 19].
1002 This interface file is an alternative interface for
1003 "cpu.weight" and allows reading and setting weight using the
1004 same values used by nice(2). Because the range is smaller and
1005 granularity is coarser for the nice values, the read value is
1006 the closest approximation of the current weight.
1009 A read-write two value file which exists on non-root cgroups.
1010 The default is "max 100000".
1012 The maximum bandwidth limit. It's in the following format::
1016 which indicates that the group may consume upto $MAX in each
1017 $PERIOD duration. "max" for $MAX indicates no limit. If only
1018 one number is written, $MAX is updated.
1021 A read-only nested-key file which exists on non-root cgroups.
1023 Shows pressure stall information for CPU. See
1024 Documentation/accounting/psi.rst for details.
1027 A read-write single value file which exists on non-root cgroups.
1028 The default is "0", i.e. no utilization boosting.
1030 The requested minimum utilization (protection) as a percentage
1031 rational number, e.g. 12.34 for 12.34%.
1033 This interface allows reading and setting minimum utilization clamp
1034 values similar to the sched_setattr(2). This minimum utilization
1035 value is used to clamp the task specific minimum utilization clamp.
1037 The requested minimum utilization (protection) is always capped by
1038 the current value for the maximum utilization (limit), i.e.
1042 A read-write single value file which exists on non-root cgroups.
1043 The default is "max". i.e. no utilization capping
1045 The requested maximum utilization (limit) as a percentage rational
1046 number, e.g. 98.76 for 98.76%.
1048 This interface allows reading and setting maximum utilization clamp
1049 values similar to the sched_setattr(2). This maximum utilization
1050 value is used to clamp the task specific maximum utilization clamp.
1057 The "memory" controller regulates distribution of memory. Memory is
1058 stateful and implements both limit and protection models. Due to the
1059 intertwining between memory usage and reclaim pressure and the
1060 stateful nature of memory, the distribution model is relatively
1063 While not completely water-tight, all major memory usages by a given
1064 cgroup are tracked so that the total memory consumption can be
1065 accounted and controlled to a reasonable extent. Currently, the
1066 following types of memory usages are tracked.
1068 - Userland memory - page cache and anonymous memory.
1070 - Kernel data structures such as dentries and inodes.
1072 - TCP socket buffers.
1074 The above list may expand in the future for better coverage.
1077 Memory Interface Files
1078 ~~~~~~~~~~~~~~~~~~~~~~
1080 All memory amounts are in bytes. If a value which is not aligned to
1081 PAGE_SIZE is written, the value may be rounded up to the closest
1082 PAGE_SIZE multiple when read back.
1085 A read-only single value file which exists on non-root
1088 The total amount of memory currently being used by the cgroup
1089 and its descendants.
1092 A read-write single value file which exists on non-root
1093 cgroups. The default is "0".
1095 Hard memory protection. If the memory usage of a cgroup
1096 is within its effective min boundary, the cgroup's memory
1097 won't be reclaimed under any conditions. If there is no
1098 unprotected reclaimable memory available, OOM killer
1099 is invoked. Above the effective min boundary (or
1100 effective low boundary if it is higher), pages are reclaimed
1101 proportionally to the overage, reducing reclaim pressure for
1104 Effective min boundary is limited by memory.min values of
1105 all ancestor cgroups. If there is memory.min overcommitment
1106 (child cgroup or cgroups are requiring more protected memory
1107 than parent will allow), then each child cgroup will get
1108 the part of parent's protection proportional to its
1109 actual memory usage below memory.min.
1111 Putting more memory than generally available under this
1112 protection is discouraged and may lead to constant OOMs.
1114 If a memory cgroup is not populated with processes,
1115 its memory.min is ignored.
1118 A read-write single value file which exists on non-root
1119 cgroups. The default is "0".
1121 Best-effort memory protection. If the memory usage of a
1122 cgroup is within its effective low boundary, the cgroup's
1123 memory won't be reclaimed unless memory can be reclaimed
1124 from unprotected cgroups. Above the effective low boundary (or
1125 effective min boundary if it is higher), pages are reclaimed
1126 proportionally to the overage, reducing reclaim pressure for
1129 Effective low boundary is limited by memory.low values of
1130 all ancestor cgroups. If there is memory.low overcommitment
1131 (child cgroup or cgroups are requiring more protected memory
1132 than parent will allow), then each child cgroup will get
1133 the part of parent's protection proportional to its
1134 actual memory usage below memory.low.
1136 Putting more memory than generally available under this
1137 protection is discouraged.
1140 A read-write single value file which exists on non-root
1141 cgroups. The default is "max".
1143 Memory usage throttle limit. This is the main mechanism to
1144 control memory usage of a cgroup. If a cgroup's usage goes
1145 over the high boundary, the processes of the cgroup are
1146 throttled and put under heavy reclaim pressure.
1148 Going over the high limit never invokes the OOM killer and
1149 under extreme conditions the limit may be breached.
1152 A read-write single value file which exists on non-root
1153 cgroups. The default is "max".
1155 Memory usage hard limit. This is the final protection
1156 mechanism. If a cgroup's memory usage reaches this limit and
1157 can't be reduced, the OOM killer is invoked in the cgroup.
1158 Under certain circumstances, the usage may go over the limit
1161 This is the ultimate protection mechanism. As long as the
1162 high limit is used and monitored properly, this limit's
1163 utility is limited to providing the final safety net.
1166 A read-write single value file which exists on non-root
1167 cgroups. The default value is "0".
1169 Determines whether the cgroup should be treated as
1170 an indivisible workload by the OOM killer. If set,
1171 all tasks belonging to the cgroup or to its descendants
1172 (if the memory cgroup is not a leaf cgroup) are killed
1173 together or not at all. This can be used to avoid
1174 partial kills to guarantee workload integrity.
1176 Tasks with the OOM protection (oom_score_adj set to -1000)
1177 are treated as an exception and are never killed.
1179 If the OOM killer is invoked in a cgroup, it's not going
1180 to kill any tasks outside of this cgroup, regardless
1181 memory.oom.group values of ancestor cgroups.
1184 A read-only flat-keyed file which exists on non-root cgroups.
1185 The following entries are defined. Unless specified
1186 otherwise, a value change in this file generates a file
1189 Note that all fields in this file are hierarchical and the
1190 file modified event can be generated due to an event down the
1191 hierarchy. For for the local events at the cgroup level see
1192 memory.events.local.
1195 The number of times the cgroup is reclaimed due to
1196 high memory pressure even though its usage is under
1197 the low boundary. This usually indicates that the low
1198 boundary is over-committed.
1201 The number of times processes of the cgroup are
1202 throttled and routed to perform direct memory reclaim
1203 because the high memory boundary was exceeded. For a
1204 cgroup whose memory usage is capped by the high limit
1205 rather than global memory pressure, this event's
1206 occurrences are expected.
1209 The number of times the cgroup's memory usage was
1210 about to go over the max boundary. If direct reclaim
1211 fails to bring it down, the cgroup goes to OOM state.
1214 The number of time the cgroup's memory usage was
1215 reached the limit and allocation was about to fail.
1217 Depending on context result could be invocation of OOM
1218 killer and retrying allocation or failing allocation.
1220 Failed allocation in its turn could be returned into
1221 userspace as -ENOMEM or silently ignored in cases like
1222 disk readahead. For now OOM in memory cgroup kills
1223 tasks iff shortage has happened inside page fault.
1225 This event is not raised if the OOM killer is not
1226 considered as an option, e.g. for failed high-order
1230 The number of processes belonging to this cgroup
1231 killed by any kind of OOM killer.
1234 Similar to memory.events but the fields in the file are local
1235 to the cgroup i.e. not hierarchical. The file modified event
1236 generated on this file reflects only the local events.
1239 A read-only flat-keyed file which exists on non-root cgroups.
1241 This breaks down the cgroup's memory footprint into different
1242 types of memory, type-specific details, and other information
1243 on the state and past events of the memory management system.
1245 All memory amounts are in bytes.
1247 The entries are ordered to be human readable, and new entries
1248 can show up in the middle. Don't rely on items remaining in a
1249 fixed position; use the keys to look up specific values!
1252 Amount of memory used in anonymous mappings such as
1253 brk(), sbrk(), and mmap(MAP_ANONYMOUS)
1256 Amount of memory used to cache filesystem data,
1257 including tmpfs and shared memory.
1260 Amount of memory allocated to kernel stacks.
1263 Amount of memory used for storing in-kernel data
1267 Amount of memory used in network transmission buffers
1270 Amount of cached filesystem data that is swap-backed,
1271 such as tmpfs, shm segments, shared anonymous mmap()s
1274 Amount of cached filesystem data mapped with mmap()
1277 Amount of cached filesystem data that was modified but
1278 not yet written back to disk
1281 Amount of cached filesystem data that was modified and
1282 is currently being written back to disk
1285 Amount of memory used in anonymous mappings backed by
1286 transparent hugepages
1288 inactive_anon, active_anon, inactive_file, active_file, unevictable
1289 Amount of memory, swap-backed and filesystem-backed,
1290 on the internal memory management lists used by the
1291 page reclaim algorithm.
1293 As these represent internal list state (eg. shmem pages are on anon
1294 memory management lists), inactive_foo + active_foo may not be equal to
1295 the value for the foo counter, since the foo counter is type-based, not
1299 Part of "slab" that might be reclaimed, such as
1300 dentries and inodes.
1303 Part of "slab" that cannot be reclaimed on memory
1307 Total number of page faults incurred
1310 Number of major page faults incurred
1314 Number of refaults of previously evicted pages
1318 Number of refaulted pages that were immediately activated
1320 workingset_nodereclaim
1322 Number of times a shadow node has been reclaimed
1326 Amount of scanned pages (in an active LRU list)
1330 Amount of scanned pages (in an inactive LRU list)
1334 Amount of reclaimed pages
1338 Amount of pages moved to the active LRU list
1342 Amount of pages moved to the inactive LRU list
1346 Amount of pages postponed to be freed under memory pressure
1350 Amount of reclaimed lazyfree pages
1354 Number of transparent hugepages which were allocated to satisfy
1355 a page fault, including COW faults. This counter is not present
1356 when CONFIG_TRANSPARENT_HUGEPAGE is not set.
1360 Number of transparent hugepages which were allocated to allow
1361 collapsing an existing range of pages. This counter is not
1362 present when CONFIG_TRANSPARENT_HUGEPAGE is not set.
1365 A read-only single value file which exists on non-root
1368 The total amount of swap currently being used by the cgroup
1369 and its descendants.
1372 A read-write single value file which exists on non-root
1373 cgroups. The default is "max".
1375 Swap usage hard limit. If a cgroup's swap usage reaches this
1376 limit, anonymous memory of the cgroup will not be swapped out.
1379 A read-only flat-keyed file which exists on non-root cgroups.
1380 The following entries are defined. Unless specified
1381 otherwise, a value change in this file generates a file
1385 The number of times the cgroup's swap usage was about
1386 to go over the max boundary and swap allocation
1390 The number of times swap allocation failed either
1391 because of running out of swap system-wide or max
1394 When reduced under the current usage, the existing swap
1395 entries are reclaimed gradually and the swap usage may stay
1396 higher than the limit for an extended period of time. This
1397 reduces the impact on the workload and memory management.
1400 A read-only nested-key file which exists on non-root cgroups.
1402 Shows pressure stall information for memory. See
1403 Documentation/accounting/psi.rst for details.
1409 "memory.high" is the main mechanism to control memory usage.
1410 Over-committing on high limit (sum of high limits > available memory)
1411 and letting global memory pressure to distribute memory according to
1412 usage is a viable strategy.
1414 Because breach of the high limit doesn't trigger the OOM killer but
1415 throttles the offending cgroup, a management agent has ample
1416 opportunities to monitor and take appropriate actions such as granting
1417 more memory or terminating the workload.
1419 Determining whether a cgroup has enough memory is not trivial as
1420 memory usage doesn't indicate whether the workload can benefit from
1421 more memory. For example, a workload which writes data received from
1422 network to a file can use all available memory but can also operate as
1423 performant with a small amount of memory. A measure of memory
1424 pressure - how much the workload is being impacted due to lack of
1425 memory - is necessary to determine whether a workload needs more
1426 memory; unfortunately, memory pressure monitoring mechanism isn't
1433 A memory area is charged to the cgroup which instantiated it and stays
1434 charged to the cgroup until the area is released. Migrating a process
1435 to a different cgroup doesn't move the memory usages that it
1436 instantiated while in the previous cgroup to the new cgroup.
1438 A memory area may be used by processes belonging to different cgroups.
1439 To which cgroup the area will be charged is in-deterministic; however,
1440 over time, the memory area is likely to end up in a cgroup which has
1441 enough memory allowance to avoid high reclaim pressure.
1443 If a cgroup sweeps a considerable amount of memory which is expected
1444 to be accessed repeatedly by other cgroups, it may make sense to use
1445 POSIX_FADV_DONTNEED to relinquish the ownership of memory areas
1446 belonging to the affected files to ensure correct memory ownership.
1452 The "io" controller regulates the distribution of IO resources. This
1453 controller implements both weight based and absolute bandwidth or IOPS
1454 limit distribution; however, weight based distribution is available
1455 only if cfq-iosched is in use and neither scheme is available for
1463 A read-only nested-keyed file which exists on non-root
1466 Lines are keyed by $MAJ:$MIN device numbers and not ordered.
1467 The following nested keys are defined.
1469 ====== =====================
1471 wbytes Bytes written
1472 rios Number of read IOs
1473 wios Number of write IOs
1474 dbytes Bytes discarded
1475 dios Number of discard IOs
1476 ====== =====================
1478 An example read output follows:
1480 8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
1481 8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
1484 A read-write nested-keyed file with exists only on the root
1487 This file configures the Quality of Service of the IO cost
1488 model based controller (CONFIG_BLK_CGROUP_IOCOST) which
1489 currently implements "io.weight" proportional control. Lines
1490 are keyed by $MAJ:$MIN device numbers and not ordered. The
1491 line for a given device is populated on the first write for
1492 the device on "io.cost.qos" or "io.cost.model". The following
1493 nested keys are defined.
1495 ====== =====================================
1496 enable Weight-based control enable
1497 ctrl "auto" or "user"
1498 rpct Read latency percentile [0, 100]
1499 rlat Read latency threshold
1500 wpct Write latency percentile [0, 100]
1501 wlat Write latency threshold
1502 min Minimum scaling percentage [1, 10000]
1503 max Maximum scaling percentage [1, 10000]
1504 ====== =====================================
1506 The controller is disabled by default and can be enabled by
1507 setting "enable" to 1. "rpct" and "wpct" parameters default
1508 to zero and the controller uses internal device saturation
1509 state to adjust the overall IO rate between "min" and "max".
1511 When a better control quality is needed, latency QoS
1512 parameters can be configured. For example::
1514 8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
1516 shows that on sdb, the controller is enabled, will consider
1517 the device saturated if the 95th percentile of read completion
1518 latencies is above 75ms or write 150ms, and adjust the overall
1519 IO issue rate between 50% and 150% accordingly.
1521 The lower the saturation point, the better the latency QoS at
1522 the cost of aggregate bandwidth. The narrower the allowed
1523 adjustment range between "min" and "max", the more conformant
1524 to the cost model the IO behavior. Note that the IO issue
1525 base rate may be far off from 100% and setting "min" and "max"
1526 blindly can lead to a significant loss of device capacity or
1527 control quality. "min" and "max" are useful for regulating
1528 devices which show wide temporary behavior changes - e.g. a
1529 ssd which accepts writes at the line speed for a while and
1530 then completely stalls for multiple seconds.
1532 When "ctrl" is "auto", the parameters are controlled by the
1533 kernel and may change automatically. Setting "ctrl" to "user"
1534 or setting any of the percentile and latency parameters puts
1535 it into "user" mode and disables the automatic changes. The
1536 automatic mode can be restored by setting "ctrl" to "auto".
1539 A read-write nested-keyed file with exists only on the root
1542 This file configures the cost model of the IO cost model based
1543 controller (CONFIG_BLK_CGROUP_IOCOST) which currently
1544 implements "io.weight" proportional control. Lines are keyed
1545 by $MAJ:$MIN device numbers and not ordered. The line for a
1546 given device is populated on the first write for the device on
1547 "io.cost.qos" or "io.cost.model". The following nested keys
1550 ===== ================================
1551 ctrl "auto" or "user"
1552 model The cost model in use - "linear"
1553 ===== ================================
1555 When "ctrl" is "auto", the kernel may change all parameters
1556 dynamically. When "ctrl" is set to "user" or any other
1557 parameters are written to, "ctrl" become "user" and the
1558 automatic changes are disabled.
1560 When "model" is "linear", the following model parameters are
1563 ============= ========================================
1564 [r|w]bps The maximum sequential IO throughput
1565 [r|w]seqiops The maximum 4k sequential IOs per second
1566 [r|w]randiops The maximum 4k random IOs per second
1567 ============= ========================================
1569 From the above, the builtin linear model determines the base
1570 costs of a sequential and random IO and the cost coefficient
1571 for the IO size. While simple, this model can cover most
1572 common device classes acceptably.
1574 The IO cost model isn't expected to be accurate in absolute
1575 sense and is scaled to the device behavior dynamically.
1577 If needed, tools/cgroup/iocost_coef_gen.py can be used to
1578 generate device-specific coefficients.
1581 A read-write flat-keyed file which exists on non-root cgroups.
1582 The default is "default 100".
1584 The first line is the default weight applied to devices
1585 without specific override. The rest are overrides keyed by
1586 $MAJ:$MIN device numbers and not ordered. The weights are in
1587 the range [1, 10000] and specifies the relative amount IO time
1588 the cgroup can use in relation to its siblings.
1590 The default weight can be updated by writing either "default
1591 $WEIGHT" or simply "$WEIGHT". Overrides can be set by writing
1592 "$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default".
1594 An example read output follows::
1601 A read-write nested-keyed file which exists on non-root
1604 BPS and IOPS based IO limit. Lines are keyed by $MAJ:$MIN
1605 device numbers and not ordered. The following nested keys are
1608 ===== ==================================
1609 rbps Max read bytes per second
1610 wbps Max write bytes per second
1611 riops Max read IO operations per second
1612 wiops Max write IO operations per second
1613 ===== ==================================
1615 When writing, any number of nested key-value pairs can be
1616 specified in any order. "max" can be specified as the value
1617 to remove a specific limit. If the same key is specified
1618 multiple times, the outcome is undefined.
1620 BPS and IOPS are measured in each IO direction and IOs are
1621 delayed if limit is reached. Temporary bursts are allowed.
1623 Setting read limit at 2M BPS and write at 120 IOPS for 8:16::
1625 echo "8:16 rbps=2097152 wiops=120" > io.max
1627 Reading returns the following::
1629 8:16 rbps=2097152 wbps=max riops=max wiops=120
1631 Write IOPS limit can be removed by writing the following::
1633 echo "8:16 wiops=max" > io.max
1635 Reading now returns the following::
1637 8:16 rbps=2097152 wbps=max riops=max wiops=max
1640 A read-only nested-key file which exists on non-root cgroups.
1642 Shows pressure stall information for IO. See
1643 Documentation/accounting/psi.rst for details.
1649 Page cache is dirtied through buffered writes and shared mmaps and
1650 written asynchronously to the backing filesystem by the writeback
1651 mechanism. Writeback sits between the memory and IO domains and
1652 regulates the proportion of dirty memory by balancing dirtying and
1655 The io controller, in conjunction with the memory controller,
1656 implements control of page cache writeback IOs. The memory controller
1657 defines the memory domain that dirty memory ratio is calculated and
1658 maintained for and the io controller defines the io domain which
1659 writes out dirty pages for the memory domain. Both system-wide and
1660 per-cgroup dirty memory states are examined and the more restrictive
1661 of the two is enforced.
1663 cgroup writeback requires explicit support from the underlying
1664 filesystem. Currently, cgroup writeback is implemented on ext2, ext4
1665 and btrfs. On other filesystems, all writeback IOs are attributed to
1668 There are inherent differences in memory and writeback management
1669 which affects how cgroup ownership is tracked. Memory is tracked per
1670 page while writeback per inode. For the purpose of writeback, an
1671 inode is assigned to a cgroup and all IO requests to write dirty pages
1672 from the inode are attributed to that cgroup.
1674 As cgroup ownership for memory is tracked per page, there can be pages
1675 which are associated with different cgroups than the one the inode is
1676 associated with. These are called foreign pages. The writeback
1677 constantly keeps track of foreign pages and, if a particular foreign
1678 cgroup becomes the majority over a certain period of time, switches
1679 the ownership of the inode to that cgroup.
1681 While this model is enough for most use cases where a given inode is
1682 mostly dirtied by a single cgroup even when the main writing cgroup
1683 changes over time, use cases where multiple cgroups write to a single
1684 inode simultaneously are not supported well. In such circumstances, a
1685 significant portion of IOs are likely to be attributed incorrectly.
1686 As memory controller assigns page ownership on the first use and
1687 doesn't update it until the page is released, even if writeback
1688 strictly follows page ownership, multiple cgroups dirtying overlapping
1689 areas wouldn't work as expected. It's recommended to avoid such usage
1692 The sysctl knobs which affect writeback behavior are applied to cgroup
1693 writeback as follows.
1695 vm.dirty_background_ratio, vm.dirty_ratio
1696 These ratios apply the same to cgroup writeback with the
1697 amount of available memory capped by limits imposed by the
1698 memory controller and system-wide clean memory.
1700 vm.dirty_background_bytes, vm.dirty_bytes
1701 For cgroup writeback, this is calculated into ratio against
1702 total available memory and applied the same way as
1703 vm.dirty[_background]_ratio.
1709 This is a cgroup v2 controller for IO workload protection. You provide a group
1710 with a latency target, and if the average latency exceeds that target the
1711 controller will throttle any peers that have a lower latency target than the
1714 The limits are only applied at the peer level in the hierarchy. This means that
1715 in the diagram below, only groups A, B, and C will influence each other, and
1716 groups D and F will influence each other. Group G will influence nobody::
1725 So the ideal way to configure this is to set io.latency in groups A, B, and C.
1726 Generally you do not want to set a value lower than the latency your device
1727 supports. Experiment to find the value that works best for your workload.
1728 Start at higher than the expected latency for your device and watch the
1729 avg_lat value in io.stat for your workload group to get an idea of the
1730 latency you see during normal operation. Use the avg_lat value as a basis for
1731 your real setting, setting at 10-15% higher than the value in io.stat.
1733 How IO Latency Throttling Works
1734 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1736 io.latency is work conserving; so as long as everybody is meeting their latency
1737 target the controller doesn't do anything. Once a group starts missing its
1738 target it begins throttling any peer group that has a higher target than itself.
1739 This throttling takes 2 forms:
1741 - Queue depth throttling. This is the number of outstanding IO's a group is
1742 allowed to have. We will clamp down relatively quickly, starting at no limit
1743 and going all the way down to 1 IO at a time.
1745 - Artificial delay induction. There are certain types of IO that cannot be
1746 throttled without possibly adversely affecting higher priority groups. This
1747 includes swapping and metadata IO. These types of IO are allowed to occur
1748 normally, however they are "charged" to the originating group. If the
1749 originating group is being throttled you will see the use_delay and delay
1750 fields in io.stat increase. The delay value is how many microseconds that are
1751 being added to any process that runs in this group. Because this number can
1752 grow quite large if there is a lot of swapping or metadata IO occurring we
1753 limit the individual delay events to 1 second at a time.
1755 Once the victimized group starts meeting its latency target again it will start
1756 unthrottling any peer groups that were throttled previously. If the victimized
1757 group simply stops doing IO the global counter will unthrottle appropriately.
1759 IO Latency Interface Files
1760 ~~~~~~~~~~~~~~~~~~~~~~~~~~
1763 This takes a similar format as the other controllers.
1765 "MAJOR:MINOR target=<target time in microseconds"
1768 If the controller is enabled you will see extra stats in io.stat in
1769 addition to the normal ones.
1772 This is the current queue depth for the group.
1775 This is an exponential moving average with a decay rate of 1/exp
1776 bound by the sampling interval. The decay rate interval can be
1777 calculated by multiplying the win value in io.stat by the
1778 corresponding number of samples based on the win value.
1781 The sampling window size in milliseconds. This is the minimum
1782 duration of time between evaluation events. Windows only elapse
1783 with IO activity. Idle periods extend the most recent window.
1788 The process number controller is used to allow a cgroup to stop any
1789 new tasks from being fork()'d or clone()'d after a specified limit is
1792 The number of tasks in a cgroup can be exhausted in ways which other
1793 controllers cannot prevent, thus warranting its own controller. For
1794 example, a fork bomb is likely to exhaust the number of tasks before
1795 hitting memory restrictions.
1797 Note that PIDs used in this controller refer to TIDs, process IDs as
1805 A read-write single value file which exists on non-root
1806 cgroups. The default is "max".
1808 Hard limit of number of processes.
1811 A read-only single value file which exists on all cgroups.
1813 The number of processes currently in the cgroup and its
1816 Organisational operations are not blocked by cgroup policies, so it is
1817 possible to have pids.current > pids.max. This can be done by either
1818 setting the limit to be smaller than pids.current, or attaching enough
1819 processes to the cgroup such that pids.current is larger than
1820 pids.max. However, it is not possible to violate a cgroup PID policy
1821 through fork() or clone(). These will return -EAGAIN if the creation
1822 of a new process would cause a cgroup policy to be violated.
1828 The "cpuset" controller provides a mechanism for constraining
1829 the CPU and memory node placement of tasks to only the resources
1830 specified in the cpuset interface files in a task's current cgroup.
1831 This is especially valuable on large NUMA systems where placing jobs
1832 on properly sized subsets of the systems with careful processor and
1833 memory placement to reduce cross-node memory access and contention
1834 can improve overall system performance.
1836 The "cpuset" controller is hierarchical. That means the controller
1837 cannot use CPUs or memory nodes not allowed in its parent.
1840 Cpuset Interface Files
1841 ~~~~~~~~~~~~~~~~~~~~~~
1844 A read-write multiple values file which exists on non-root
1845 cpuset-enabled cgroups.
1847 It lists the requested CPUs to be used by tasks within this
1848 cgroup. The actual list of CPUs to be granted, however, is
1849 subjected to constraints imposed by its parent and can differ
1850 from the requested CPUs.
1852 The CPU numbers are comma-separated numbers or ranges.
1858 An empty value indicates that the cgroup is using the same
1859 setting as the nearest cgroup ancestor with a non-empty
1860 "cpuset.cpus" or all the available CPUs if none is found.
1862 The value of "cpuset.cpus" stays constant until the next update
1863 and won't be affected by any CPU hotplug events.
1865 cpuset.cpus.effective
1866 A read-only multiple values file which exists on all
1867 cpuset-enabled cgroups.
1869 It lists the onlined CPUs that are actually granted to this
1870 cgroup by its parent. These CPUs are allowed to be used by
1871 tasks within the current cgroup.
1873 If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows
1874 all the CPUs from the parent cgroup that can be available to
1875 be used by this cgroup. Otherwise, it should be a subset of
1876 "cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus"
1877 can be granted. In this case, it will be treated just like an
1878 empty "cpuset.cpus".
1880 Its value will be affected by CPU hotplug events.
1883 A read-write multiple values file which exists on non-root
1884 cpuset-enabled cgroups.
1886 It lists the requested memory nodes to be used by tasks within
1887 this cgroup. The actual list of memory nodes granted, however,
1888 is subjected to constraints imposed by its parent and can differ
1889 from the requested memory nodes.
1891 The memory node numbers are comma-separated numbers or ranges.
1897 An empty value indicates that the cgroup is using the same
1898 setting as the nearest cgroup ancestor with a non-empty
1899 "cpuset.mems" or all the available memory nodes if none
1902 The value of "cpuset.mems" stays constant until the next update
1903 and won't be affected by any memory nodes hotplug events.
1905 cpuset.mems.effective
1906 A read-only multiple values file which exists on all
1907 cpuset-enabled cgroups.
1909 It lists the onlined memory nodes that are actually granted to
1910 this cgroup by its parent. These memory nodes are allowed to
1911 be used by tasks within the current cgroup.
1913 If "cpuset.mems" is empty, it shows all the memory nodes from the
1914 parent cgroup that will be available to be used by this cgroup.
1915 Otherwise, it should be a subset of "cpuset.mems" unless none of
1916 the memory nodes listed in "cpuset.mems" can be granted. In this
1917 case, it will be treated just like an empty "cpuset.mems".
1919 Its value will be affected by memory nodes hotplug events.
1921 cpuset.cpus.partition
1922 A read-write single value file which exists on non-root
1923 cpuset-enabled cgroups. This flag is owned by the parent cgroup
1924 and is not delegatable.
1926 It accepts only the following input values when written to.
1928 "root" - a paritition root
1929 "member" - a non-root member of a partition
1931 When set to be a partition root, the current cgroup is the
1932 root of a new partition or scheduling domain that comprises
1933 itself and all its descendants except those that are separate
1934 partition roots themselves and their descendants. The root
1935 cgroup is always a partition root.
1937 There are constraints on where a partition root can be set.
1938 It can only be set in a cgroup if all the following conditions
1941 1) The "cpuset.cpus" is not empty and the list of CPUs are
1942 exclusive, i.e. they are not shared by any of its siblings.
1943 2) The parent cgroup is a partition root.
1944 3) The "cpuset.cpus" is also a proper subset of the parent's
1945 "cpuset.cpus.effective".
1946 4) There is no child cgroups with cpuset enabled. This is for
1947 eliminating corner cases that have to be handled if such a
1948 condition is allowed.
1950 Setting it to partition root will take the CPUs away from the
1951 effective CPUs of the parent cgroup. Once it is set, this
1952 file cannot be reverted back to "member" if there are any child
1953 cgroups with cpuset enabled.
1955 A parent partition cannot distribute all its CPUs to its
1956 child partitions. There must be at least one cpu left in the
1959 Once becoming a partition root, changes to "cpuset.cpus" is
1960 generally allowed as long as the first condition above is true,
1961 the change will not take away all the CPUs from the parent
1962 partition and the new "cpuset.cpus" value is a superset of its
1963 children's "cpuset.cpus" values.
1965 Sometimes, external factors like changes to ancestors'
1966 "cpuset.cpus" or cpu hotplug can cause the state of the partition
1967 root to change. On read, the "cpuset.sched.partition" file
1968 can show the following values.
1970 "member" Non-root member of a partition
1971 "root" Partition root
1972 "root invalid" Invalid partition root
1974 It is a partition root if the first 2 partition root conditions
1975 above are true and at least one CPU from "cpuset.cpus" is
1976 granted by the parent cgroup.
1978 A partition root can become invalid if none of CPUs requested
1979 in "cpuset.cpus" can be granted by the parent cgroup or the
1980 parent cgroup is no longer a partition root itself. In this
1981 case, it is not a real partition even though the restriction
1982 of the first partition root condition above will still apply.
1983 The cpu affinity of all the tasks in the cgroup will then be
1984 associated with CPUs in the nearest ancestor partition.
1986 An invalid partition root can be transitioned back to a
1987 real partition root if at least one of the requested CPUs
1988 can now be granted by its parent. In this case, the cpu
1989 affinity of all the tasks in the formerly invalid partition
1990 will be associated to the CPUs of the newly formed partition.
1991 Changing the partition state of an invalid partition root to
1992 "member" is always allowed even if child cpusets are present.
1998 Device controller manages access to device files. It includes both
1999 creation of new device files (using mknod), and access to the
2000 existing device files.
2002 Cgroup v2 device controller has no interface files and is implemented
2003 on top of cgroup BPF. To control access to device files, a user may
2004 create bpf programs of the BPF_CGROUP_DEVICE type and attach them
2005 to cgroups. On an attempt to access a device file, corresponding
2006 BPF programs will be executed, and depending on the return value
2007 the attempt will succeed or fail with -EPERM.
2009 A BPF_CGROUP_DEVICE program takes a pointer to the bpf_cgroup_dev_ctx
2010 structure, which describes the device access attempt: access type
2011 (mknod/read/write) and device (type, major and minor numbers).
2012 If the program returns 0, the attempt fails with -EPERM, otherwise
2015 An example of BPF_CGROUP_DEVICE program may be found in the kernel
2016 source tree in the tools/testing/selftests/bpf/dev_cgroup.c file.
2022 The "rdma" controller regulates the distribution and accounting of
2025 RDMA Interface Files
2026 ~~~~~~~~~~~~~~~~~~~~
2029 A readwrite nested-keyed file that exists for all the cgroups
2030 except root that describes current configured resource limit
2031 for a RDMA/IB device.
2033 Lines are keyed by device name and are not ordered.
2034 Each line contains space separated resource name and its configured
2035 limit that can be distributed.
2037 The following nested keys are defined.
2039 ========== =============================
2040 hca_handle Maximum number of HCA Handles
2041 hca_object Maximum number of HCA Objects
2042 ========== =============================
2044 An example for mlx4 and ocrdma device follows::
2046 mlx4_0 hca_handle=2 hca_object=2000
2047 ocrdma1 hca_handle=3 hca_object=max
2050 A read-only file that describes current resource usage.
2051 It exists for all the cgroup except root.
2053 An example for mlx4 and ocrdma device follows::
2055 mlx4_0 hca_handle=1 hca_object=20
2056 ocrdma1 hca_handle=1 hca_object=23
2065 perf_event controller, if not mounted on a legacy hierarchy, is
2066 automatically enabled on the v2 hierarchy so that perf events can
2067 always be filtered by cgroup v2 path. The controller can still be
2068 moved to a legacy hierarchy after v2 hierarchy is populated.
2071 Non-normative information
2072 -------------------------
2074 This section contains information that isn't considered to be a part of
2075 the stable kernel API and so is subject to change.
2078 CPU controller root cgroup process behaviour
2079 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2081 When distributing CPU cycles in the root cgroup each thread in this
2082 cgroup is treated as if it was hosted in a separate child cgroup of the
2083 root cgroup. This child cgroup weight is dependent on its thread nice
2086 For details of this mapping see sched_prio_to_weight array in
2087 kernel/sched/core.c file (values from this array should be scaled
2088 appropriately so the neutral - nice 0 - value is 100 instead of 1024).
2091 IO controller root cgroup process behaviour
2092 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2094 Root cgroup processes are hosted in an implicit leaf child node.
2095 When distributing IO resources this implicit child node is taken into
2096 account as if it was a normal child cgroup of the root cgroup with a
2097 weight value of 200.
2106 cgroup namespace provides a mechanism to virtualize the view of the
2107 "/proc/$PID/cgroup" file and cgroup mounts. The CLONE_NEWCGROUP clone
2108 flag can be used with clone(2) and unshare(2) to create a new cgroup
2109 namespace. The process running inside the cgroup namespace will have
2110 its "/proc/$PID/cgroup" output restricted to cgroupns root. The
2111 cgroupns root is the cgroup of the process at the time of creation of
2112 the cgroup namespace.
2114 Without cgroup namespace, the "/proc/$PID/cgroup" file shows the
2115 complete path of the cgroup of a process. In a container setup where
2116 a set of cgroups and namespaces are intended to isolate processes the
2117 "/proc/$PID/cgroup" file may leak potential system level information
2118 to the isolated processes. For Example::
2120 # cat /proc/self/cgroup
2121 0::/batchjobs/container_id1
2123 The path '/batchjobs/container_id1' can be considered as system-data
2124 and undesirable to expose to the isolated processes. cgroup namespace
2125 can be used to restrict visibility of this path. For example, before
2126 creating a cgroup namespace, one would see::
2128 # ls -l /proc/self/ns/cgroup
2129 lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835]
2130 # cat /proc/self/cgroup
2131 0::/batchjobs/container_id1
2133 After unsharing a new namespace, the view changes::
2135 # ls -l /proc/self/ns/cgroup
2136 lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183]
2137 # cat /proc/self/cgroup
2140 When some thread from a multi-threaded process unshares its cgroup
2141 namespace, the new cgroupns gets applied to the entire process (all
2142 the threads). This is natural for the v2 hierarchy; however, for the
2143 legacy hierarchies, this may be unexpected.
2145 A cgroup namespace is alive as long as there are processes inside or
2146 mounts pinning it. When the last usage goes away, the cgroup
2147 namespace is destroyed. The cgroupns root and the actual cgroups
2154 The 'cgroupns root' for a cgroup namespace is the cgroup in which the
2155 process calling unshare(2) is running. For example, if a process in
2156 /batchjobs/container_id1 cgroup calls unshare, cgroup
2157 /batchjobs/container_id1 becomes the cgroupns root. For the
2158 init_cgroup_ns, this is the real root ('/') cgroup.
2160 The cgroupns root cgroup does not change even if the namespace creator
2161 process later moves to a different cgroup::
2163 # ~/unshare -c # unshare cgroupns in some cgroup
2164 # cat /proc/self/cgroup
2167 # echo 0 > sub_cgrp_1/cgroup.procs
2168 # cat /proc/self/cgroup
2171 Each process gets its namespace-specific view of "/proc/$PID/cgroup"
2173 Processes running inside the cgroup namespace will be able to see
2174 cgroup paths (in /proc/self/cgroup) only inside their root cgroup.
2175 From within an unshared cgroupns::
2179 # echo 7353 > sub_cgrp_1/cgroup.procs
2180 # cat /proc/7353/cgroup
2183 From the initial cgroup namespace, the real cgroup path will be
2186 $ cat /proc/7353/cgroup
2187 0::/batchjobs/container_id1/sub_cgrp_1
2189 From a sibling cgroup namespace (that is, a namespace rooted at a
2190 different cgroup), the cgroup path relative to its own cgroup
2191 namespace root will be shown. For instance, if PID 7353's cgroup
2192 namespace root is at '/batchjobs/container_id2', then it will see::
2194 # cat /proc/7353/cgroup
2195 0::/../container_id2/sub_cgrp_1
2197 Note that the relative path always starts with '/' to indicate that
2198 its relative to the cgroup namespace root of the caller.
2201 Migration and setns(2)
2202 ----------------------
2204 Processes inside a cgroup namespace can move into and out of the
2205 namespace root if they have proper access to external cgroups. For
2206 example, from inside a namespace with cgroupns root at
2207 /batchjobs/container_id1, and assuming that the global hierarchy is
2208 still accessible inside cgroupns::
2210 # cat /proc/7353/cgroup
2212 # echo 7353 > batchjobs/container_id2/cgroup.procs
2213 # cat /proc/7353/cgroup
2214 0::/../container_id2
2216 Note that this kind of setup is not encouraged. A task inside cgroup
2217 namespace should only be exposed to its own cgroupns hierarchy.
2219 setns(2) to another cgroup namespace is allowed when:
2221 (a) the process has CAP_SYS_ADMIN against its current user namespace
2222 (b) the process has CAP_SYS_ADMIN against the target cgroup
2225 No implicit cgroup changes happen with attaching to another cgroup
2226 namespace. It is expected that the someone moves the attaching
2227 process under the target cgroup namespace root.
2230 Interaction with Other Namespaces
2231 ---------------------------------
2233 Namespace specific cgroup hierarchy can be mounted by a process
2234 running inside a non-init cgroup namespace::
2236 # mount -t cgroup2 none $MOUNT_POINT
2238 This will mount the unified cgroup hierarchy with cgroupns root as the
2239 filesystem root. The process needs CAP_SYS_ADMIN against its user and
2242 The virtualization of /proc/self/cgroup file combined with restricting
2243 the view of cgroup hierarchy by namespace-private cgroupfs mount
2244 provides a properly isolated cgroup view inside the container.
2247 Information on Kernel Programming
2248 =================================
2250 This section contains kernel programming information in the areas
2251 where interacting with cgroup is necessary. cgroup core and
2252 controllers are not covered.
2255 Filesystem Support for Writeback
2256 --------------------------------
2258 A filesystem can support cgroup writeback by updating
2259 address_space_operations->writepage[s]() to annotate bio's using the
2260 following two functions.
2262 wbc_init_bio(@wbc, @bio)
2263 Should be called for each bio carrying writeback data and
2264 associates the bio with the inode's owner cgroup and the
2265 corresponding request queue. This must be called after
2266 a queue (device) has been associated with the bio and
2269 wbc_account_cgroup_owner(@wbc, @page, @bytes)
2270 Should be called for each data segment being written out.
2271 While this function doesn't care exactly when it's called
2272 during the writeback session, it's the easiest and most
2273 natural to call it as data segments are added to a bio.
2275 With writeback bio's annotated, cgroup support can be enabled per
2276 super_block by setting SB_I_CGROUPWB in ->s_iflags. This allows for
2277 selective disabling of cgroup writeback support which is helpful when
2278 certain filesystem features, e.g. journaled data mode, are
2281 wbc_init_bio() binds the specified bio to its cgroup. Depending on
2282 the configuration, the bio may be executed at a lower priority and if
2283 the writeback session is holding shared resources, e.g. a journal
2284 entry, may lead to priority inversion. There is no one easy solution
2285 for the problem. Filesystems can try to work around specific problem
2286 cases by skipping wbc_init_bio() and using bio_associate_blkg()
2290 Deprecated v1 Core Features
2291 ===========================
2293 - Multiple hierarchies including named ones are not supported.
2295 - All v1 mount options are not supported.
2297 - The "tasks" file is removed and "cgroup.procs" is not sorted.
2299 - "cgroup.clone_children" is removed.
2301 - /proc/cgroups is meaningless for v2. Use "cgroup.controllers" file
2302 at the root instead.
2305 Issues with v1 and Rationales for v2
2306 ====================================
2308 Multiple Hierarchies
2309 --------------------
2311 cgroup v1 allowed an arbitrary number of hierarchies and each
2312 hierarchy could host any number of controllers. While this seemed to
2313 provide a high level of flexibility, it wasn't useful in practice.
2315 For example, as there is only one instance of each controller, utility
2316 type controllers such as freezer which can be useful in all
2317 hierarchies could only be used in one. The issue is exacerbated by
2318 the fact that controllers couldn't be moved to another hierarchy once
2319 hierarchies were populated. Another issue was that all controllers
2320 bound to a hierarchy were forced to have exactly the same view of the
2321 hierarchy. It wasn't possible to vary the granularity depending on
2322 the specific controller.
2324 In practice, these issues heavily limited which controllers could be
2325 put on the same hierarchy and most configurations resorted to putting
2326 each controller on its own hierarchy. Only closely related ones, such
2327 as the cpu and cpuacct controllers, made sense to be put on the same
2328 hierarchy. This often meant that userland ended up managing multiple
2329 similar hierarchies repeating the same steps on each hierarchy
2330 whenever a hierarchy management operation was necessary.
2332 Furthermore, support for multiple hierarchies came at a steep cost.
2333 It greatly complicated cgroup core implementation but more importantly
2334 the support for multiple hierarchies restricted how cgroup could be
2335 used in general and what controllers was able to do.
2337 There was no limit on how many hierarchies there might be, which meant
2338 that a thread's cgroup membership couldn't be described in finite
2339 length. The key might contain any number of entries and was unlimited
2340 in length, which made it highly awkward to manipulate and led to
2341 addition of controllers which existed only to identify membership,
2342 which in turn exacerbated the original problem of proliferating number
2345 Also, as a controller couldn't have any expectation regarding the
2346 topologies of hierarchies other controllers might be on, each
2347 controller had to assume that all other controllers were attached to
2348 completely orthogonal hierarchies. This made it impossible, or at
2349 least very cumbersome, for controllers to cooperate with each other.
2351 In most use cases, putting controllers on hierarchies which are
2352 completely orthogonal to each other isn't necessary. What usually is
2353 called for is the ability to have differing levels of granularity
2354 depending on the specific controller. In other words, hierarchy may
2355 be collapsed from leaf towards root when viewed from specific
2356 controllers. For example, a given configuration might not care about
2357 how memory is distributed beyond a certain level while still wanting
2358 to control how CPU cycles are distributed.
2364 cgroup v1 allowed threads of a process to belong to different cgroups.
2365 This didn't make sense for some controllers and those controllers
2366 ended up implementing different ways to ignore such situations but
2367 much more importantly it blurred the line between API exposed to
2368 individual applications and system management interface.
2370 Generally, in-process knowledge is available only to the process
2371 itself; thus, unlike service-level organization of processes,
2372 categorizing threads of a process requires active participation from
2373 the application which owns the target process.
2375 cgroup v1 had an ambiguously defined delegation model which got abused
2376 in combination with thread granularity. cgroups were delegated to
2377 individual applications so that they can create and manage their own
2378 sub-hierarchies and control resource distributions along them. This
2379 effectively raised cgroup to the status of a syscall-like API exposed
2382 First of all, cgroup has a fundamentally inadequate interface to be
2383 exposed this way. For a process to access its own knobs, it has to
2384 extract the path on the target hierarchy from /proc/self/cgroup,
2385 construct the path by appending the name of the knob to the path, open
2386 and then read and/or write to it. This is not only extremely clunky
2387 and unusual but also inherently racy. There is no conventional way to
2388 define transaction across the required steps and nothing can guarantee
2389 that the process would actually be operating on its own sub-hierarchy.
2391 cgroup controllers implemented a number of knobs which would never be
2392 accepted as public APIs because they were just adding control knobs to
2393 system-management pseudo filesystem. cgroup ended up with interface
2394 knobs which were not properly abstracted or refined and directly
2395 revealed kernel internal details. These knobs got exposed to
2396 individual applications through the ill-defined delegation mechanism
2397 effectively abusing cgroup as a shortcut to implementing public APIs
2398 without going through the required scrutiny.
2400 This was painful for both userland and kernel. Userland ended up with
2401 misbehaving and poorly abstracted interfaces and kernel exposing and
2402 locked into constructs inadvertently.
2405 Competition Between Inner Nodes and Threads
2406 -------------------------------------------
2408 cgroup v1 allowed threads to be in any cgroups which created an
2409 interesting problem where threads belonging to a parent cgroup and its
2410 children cgroups competed for resources. This was nasty as two
2411 different types of entities competed and there was no obvious way to
2412 settle it. Different controllers did different things.
2414 The cpu controller considered threads and cgroups as equivalents and
2415 mapped nice levels to cgroup weights. This worked for some cases but
2416 fell flat when children wanted to be allocated specific ratios of CPU
2417 cycles and the number of internal threads fluctuated - the ratios
2418 constantly changed as the number of competing entities fluctuated.
2419 There also were other issues. The mapping from nice level to weight
2420 wasn't obvious or universal, and there were various other knobs which
2421 simply weren't available for threads.
2423 The io controller implicitly created a hidden leaf node for each
2424 cgroup to host the threads. The hidden leaf had its own copies of all
2425 the knobs with ``leaf_`` prefixed. While this allowed equivalent
2426 control over internal threads, it was with serious drawbacks. It
2427 always added an extra layer of nesting which wouldn't be necessary
2428 otherwise, made the interface messy and significantly complicated the
2431 The memory controller didn't have a way to control what happened
2432 between internal tasks and child cgroups and the behavior was not
2433 clearly defined. There were attempts to add ad-hoc behaviors and
2434 knobs to tailor the behavior to specific workloads which would have
2435 led to problems extremely difficult to resolve in the long term.
2437 Multiple controllers struggled with internal tasks and came up with
2438 different ways to deal with it; unfortunately, all the approaches were
2439 severely flawed and, furthermore, the widely different behaviors
2440 made cgroup as a whole highly inconsistent.
2442 This clearly is a problem which needs to be addressed from cgroup core
2446 Other Interface Issues
2447 ----------------------
2449 cgroup v1 grew without oversight and developed a large number of
2450 idiosyncrasies and inconsistencies. One issue on the cgroup core side
2451 was how an empty cgroup was notified - a userland helper binary was
2452 forked and executed for each event. The event delivery wasn't
2453 recursive or delegatable. The limitations of the mechanism also led
2454 to in-kernel event delivery filtering mechanism further complicating
2457 Controller interfaces were problematic too. An extreme example is
2458 controllers completely ignoring hierarchical organization and treating
2459 all cgroups as if they were all located directly under the root
2460 cgroup. Some controllers exposed a large amount of inconsistent
2461 implementation details to userland.
2463 There also was no consistency across controllers. When a new cgroup
2464 was created, some controllers defaulted to not imposing extra
2465 restrictions while others disallowed any resource usage until
2466 explicitly configured. Configuration knobs for the same type of
2467 control used widely differing naming schemes and formats. Statistics
2468 and information knobs were named arbitrarily and used different
2469 formats and units even in the same controller.
2471 cgroup v2 establishes common conventions where appropriate and updates
2472 controllers so that they expose minimal and consistent interfaces.
2475 Controller Issues and Remedies
2476 ------------------------------
2481 The original lower boundary, the soft limit, is defined as a limit
2482 that is per default unset. As a result, the set of cgroups that
2483 global reclaim prefers is opt-in, rather than opt-out. The costs for
2484 optimizing these mostly negative lookups are so high that the
2485 implementation, despite its enormous size, does not even provide the
2486 basic desirable behavior. First off, the soft limit has no
2487 hierarchical meaning. All configured groups are organized in a global
2488 rbtree and treated like equal peers, regardless where they are located
2489 in the hierarchy. This makes subtree delegation impossible. Second,
2490 the soft limit reclaim pass is so aggressive that it not just
2491 introduces high allocation latencies into the system, but also impacts
2492 system performance due to overreclaim, to the point where the feature
2493 becomes self-defeating.
2495 The memory.low boundary on the other hand is a top-down allocated
2496 reserve. A cgroup enjoys reclaim protection when it's within its
2497 effective low, which makes delegation of subtrees possible. It also
2498 enjoys having reclaim pressure proportional to its overage when
2499 above its effective low.
2501 The original high boundary, the hard limit, is defined as a strict
2502 limit that can not budge, even if the OOM killer has to be called.
2503 But this generally goes against the goal of making the most out of the
2504 available memory. The memory consumption of workloads varies during
2505 runtime, and that requires users to overcommit. But doing that with a
2506 strict upper limit requires either a fairly accurate prediction of the
2507 working set size or adding slack to the limit. Since working set size
2508 estimation is hard and error prone, and getting it wrong results in
2509 OOM kills, most users tend to err on the side of a looser limit and
2510 end up wasting precious resources.
2512 The memory.high boundary on the other hand can be set much more
2513 conservatively. When hit, it throttles allocations by forcing them
2514 into direct reclaim to work off the excess, but it never invokes the
2515 OOM killer. As a result, a high boundary that is chosen too
2516 aggressively will not terminate the processes, but instead it will
2517 lead to gradual performance degradation. The user can monitor this
2518 and make corrections until the minimal memory footprint that still
2519 gives acceptable performance is found.
2521 In extreme cases, with many concurrent allocations and a complete
2522 breakdown of reclaim progress within the group, the high boundary can
2523 be exceeded. But even then it's mostly better to satisfy the
2524 allocation from the slack available in other groups or the rest of the
2525 system than killing the group. Otherwise, memory.max is there to
2526 limit this type of spillover and ultimately contain buggy or even
2527 malicious applications.
2529 Setting the original memory.limit_in_bytes below the current usage was
2530 subject to a race condition, where concurrent charges could cause the
2531 limit setting to fail. memory.max on the other hand will first set the
2532 limit to prevent new charges, and then reclaim and OOM kill until the
2533 new limit is met - or the task writing to memory.max is killed.
2535 The combined memory+swap accounting and limiting is replaced by real
2536 control over swap space.
2538 The main argument for a combined memory+swap facility in the original
2539 cgroup design was that global or parental pressure would always be
2540 able to swap all anonymous memory of a child group, regardless of the
2541 child's own (possibly untrusted) configuration. However, untrusted
2542 groups can sabotage swapping by other means - such as referencing its
2543 anonymous memory in a tight loop - and an admin can not assume full
2544 swappability when overcommitting untrusted jobs.
2546 For trusted jobs, on the other hand, a combined counter is not an
2547 intuitive userspace interface, and it flies in the face of the idea
2548 that cgroup controllers should account and limit specific physical
2549 resources. Swap space is a resource like all others in the system,
2550 and that's why unified hierarchy allows distributing it separately.