1.. _cgroup-v2: 2 3================ 4Control Group v2 5================ 6 7:Date: October, 2015 8:Author: Tejun Heo <tj@kernel.org> 9 10This is the authoritative documentation on the design, interface and 11conventions of cgroup v2. It describes all userland-visible aspects 12of cgroup including core and specific controller behaviors. All 13future changes must be reflected in this document. Documentation for 14v1 is available under :ref:`Documentation/admin-guide/cgroup-v1/index.rst <cgroup-v1>`. 15 16.. CONTENTS 17 18 1. Introduction 19 1-1. Terminology 20 1-2. What is cgroup? 21 2. Basic Operations 22 2-1. Mounting 23 2-2. Organizing Processes and Threads 24 2-2-1. Processes 25 2-2-2. Threads 26 2-3. [Un]populated Notification 27 2-4. Controlling Controllers 28 2-4-1. Enabling and Disabling 29 2-4-2. Top-down Constraint 30 2-4-3. No Internal Process Constraint 31 2-5. Delegation 32 2-5-1. Model of Delegation 33 2-5-2. Delegation Containment 34 2-6. Guidelines 35 2-6-1. Organize Once and Control 36 2-6-2. Avoid Name Collisions 37 3. Resource Distribution Models 38 3-1. Weights 39 3-2. Limits 40 3-3. Protections 41 3-4. Allocations 42 4. Interface Files 43 4-1. Format 44 4-2. Conventions 45 4-3. Core Interface Files 46 5. Controllers 47 5-1. CPU 48 5-1-1. CPU Interface Files 49 5-2. Memory 50 5-2-1. Memory Interface Files 51 5-2-2. Usage Guidelines 52 5-2-3. Memory Ownership 53 5-3. IO 54 5-3-1. IO Interface Files 55 5-3-2. Writeback 56 5-3-3. IO Latency 57 5-3-3-1. How IO Latency Throttling Works 58 5-3-3-2. IO Latency Interface Files 59 5-3-4. IO Priority 60 5-4. PID 61 5-4-1. PID Interface Files 62 5-5. Cpuset 63 5.5-1. Cpuset Interface Files 64 5-6. Device 65 5-7. RDMA 66 5-7-1. RDMA Interface Files 67 5-8. HugeTLB 68 5.8-1. HugeTLB Interface Files 69 5-9. Misc 70 5.9-1 Miscellaneous cgroup Interface Files 71 5.9-2 Migration and Ownership 72 5-10. Others 73 5-10-1. perf_event 74 5-N. Non-normative information 75 5-N-1. CPU controller root cgroup process behaviour 76 5-N-2. IO controller root cgroup process behaviour 77 6. Namespace 78 6-1. Basics 79 6-2. The Root and Views 80 6-3. Migration and setns(2) 81 6-4. Interaction with Other Namespaces 82 P. Information on Kernel Programming 83 P-1. Filesystem Support for Writeback 84 D. Deprecated v1 Core Features 85 R. Issues with v1 and Rationales for v2 86 R-1. Multiple Hierarchies 87 R-2. Thread Granularity 88 R-3. Competition Between Inner Nodes and Threads 89 R-4. Other Interface Issues 90 R-5. Controller Issues and Remedies 91 R-5-1. Memory 92 93 94Introduction 95============ 96 97Terminology 98----------- 99 100"cgroup" stands for "control group" and is never capitalized. The 101singular form is used to designate the whole feature and also as a 102qualifier as in "cgroup controllers". When explicitly referring to 103multiple individual control groups, the plural form "cgroups" is used. 104 105 106What is cgroup? 107--------------- 108 109cgroup is a mechanism to organize processes hierarchically and 110distribute system resources along the hierarchy in a controlled and 111configurable manner. 112 113cgroup is largely composed of two parts - the core and controllers. 114cgroup core is primarily responsible for hierarchically organizing 115processes. A cgroup controller is usually responsible for 116distributing a specific type of system resource along the hierarchy 117although there are utility controllers which serve purposes other than 118resource distribution. 119 120cgroups form a tree structure and every process in the system belongs 121to one and only one cgroup. All threads of a process belong to the 122same cgroup. On creation, all processes are put in the cgroup that 123the parent process belongs to at the time. A process can be migrated 124to another cgroup. Migration of a process doesn't affect already 125existing descendant processes. 126 127Following certain structural constraints, controllers may be enabled or 128disabled selectively on a cgroup. All controller behaviors are 129hierarchical - if a controller is enabled on a cgroup, it affects all 130processes which belong to the cgroups consisting the inclusive 131sub-hierarchy of the cgroup. When a controller is enabled on a nested 132cgroup, it always restricts the resource distribution further. The 133restrictions set closer to the root in the hierarchy can not be 134overridden from further away. 135 136 137Basic Operations 138================ 139 140Mounting 141-------- 142 143Unlike v1, cgroup v2 has only single hierarchy. The cgroup v2 144hierarchy can be mounted with the following mount command:: 145 146 # mount -t cgroup2 none $MOUNT_POINT 147 148cgroup2 filesystem has the magic number 0x63677270 ("cgrp"). All 149controllers which support v2 and are not bound to a v1 hierarchy are 150automatically bound to the v2 hierarchy and show up at the root. 151Controllers which are not in active use in the v2 hierarchy can be 152bound to other hierarchies. This allows mixing v2 hierarchy with the 153legacy v1 multiple hierarchies in a fully backward compatible way. 154 155A controller can be moved across hierarchies only after the controller 156is no longer referenced in its current hierarchy. Because per-cgroup 157controller states are destroyed asynchronously and controllers may 158have lingering references, a controller may not show up immediately on 159the v2 hierarchy after the final umount of the previous hierarchy. 160Similarly, a controller should be fully disabled to be moved out of 161the unified hierarchy and it may take some time for the disabled 162controller to become available for other hierarchies; furthermore, due 163to inter-controller dependencies, other controllers may need to be 164disabled too. 165 166While useful for development and manual configurations, moving 167controllers dynamically between the v2 and other hierarchies is 168strongly discouraged for production use. It is recommended to decide 169the hierarchies and controller associations before starting using the 170controllers after system boot. 171 172During transition to v2, system management software might still 173automount the v1 cgroup filesystem and so hijack all controllers 174during boot, before manual intervention is possible. To make testing 175and experimenting easier, the kernel parameter cgroup_no_v1= allows 176disabling controllers in v1 and make them always available in v2. 177 178cgroup v2 currently supports the following mount options. 179 180 nsdelegate 181 Consider cgroup namespaces as delegation boundaries. This 182 option is system wide and can only be set on mount or modified 183 through remount from the init namespace. The mount option is 184 ignored on non-init namespace mounts. Please refer to the 185 Delegation section for details. 186 187 favordynmods 188 Reduce the latencies of dynamic cgroup modifications such as 189 task migrations and controller on/offs at the cost of making 190 hot path operations such as forks and exits more expensive. 191 The static usage pattern of creating a cgroup, enabling 192 controllers, and then seeding it with CLONE_INTO_CGROUP is 193 not affected by this option. 194 195 memory_localevents 196 Only populate memory.events with data for the current cgroup, 197 and not any subtrees. This is legacy behaviour, the default 198 behaviour without this option is to include subtree counts. 199 This option is system wide and can only be set on mount or 200 modified through remount from the init namespace. The mount 201 option is ignored on non-init namespace mounts. 202 203 memory_recursiveprot 204 Recursively apply memory.min and memory.low protection to 205 entire subtrees, without requiring explicit downward 206 propagation into leaf cgroups. This allows protecting entire 207 subtrees from one another, while retaining free competition 208 within those subtrees. This should have been the default 209 behavior but is a mount-option to avoid regressing setups 210 relying on the original semantics (e.g. specifying bogusly 211 high 'bypass' protection values at higher tree levels). 212 213 214Organizing Processes and Threads 215-------------------------------- 216 217Processes 218~~~~~~~~~ 219 220Initially, only the root cgroup exists to which all processes belong. 221A child cgroup can be created by creating a sub-directory:: 222 223 # mkdir $CGROUP_NAME 224 225A given cgroup may have multiple child cgroups forming a tree 226structure. Each cgroup has a read-writable interface file 227"cgroup.procs". When read, it lists the PIDs of all processes which 228belong to the cgroup one-per-line. The PIDs are not ordered and the 229same PID may show up more than once if the process got moved to 230another cgroup and then back or the PID got recycled while reading. 231 232A process can be migrated into a cgroup by writing its PID to the 233target cgroup's "cgroup.procs" file. Only one process can be migrated 234on a single write(2) call. If a process is composed of multiple 235threads, writing the PID of any thread migrates all threads of the 236process. 237 238When a process forks a child process, the new process is born into the 239cgroup that the forking process belongs to at the time of the 240operation. After exit, a process stays associated with the cgroup 241that it belonged to at the time of exit until it's reaped; however, a 242zombie process does not appear in "cgroup.procs" and thus can't be 243moved to another cgroup. 244 245A cgroup which doesn't have any children or live processes can be 246destroyed by removing the directory. Note that a cgroup which doesn't 247have any children and is associated only with zombie processes is 248considered empty and can be removed:: 249 250 # rmdir $CGROUP_NAME 251 252"/proc/$PID/cgroup" lists a process's cgroup membership. If legacy 253cgroup is in use in the system, this file may contain multiple lines, 254one for each hierarchy. The entry for cgroup v2 is always in the 255format "0::$PATH":: 256 257 # cat /proc/842/cgroup 258 ... 259 0::/test-cgroup/test-cgroup-nested 260 261If the process becomes a zombie and the cgroup it was associated with 262is removed subsequently, " (deleted)" is appended to the path:: 263 264 # cat /proc/842/cgroup 265 ... 266 0::/test-cgroup/test-cgroup-nested (deleted) 267 268 269Threads 270~~~~~~~ 271 272cgroup v2 supports thread granularity for a subset of controllers to 273support use cases requiring hierarchical resource distribution across 274the threads of a group of processes. By default, all threads of a 275process belong to the same cgroup, which also serves as the resource 276domain to host resource consumptions which are not specific to a 277process or thread. The thread mode allows threads to be spread across 278a subtree while still maintaining the common resource domain for them. 279 280Controllers which support thread mode are called threaded controllers. 281The ones which don't are called domain controllers. 282 283Marking a cgroup threaded makes it join the resource domain of its 284parent as a threaded cgroup. The parent may be another threaded 285cgroup whose resource domain is further up in the hierarchy. The root 286of a threaded subtree, that is, the nearest ancestor which is not 287threaded, is called threaded domain or thread root interchangeably and 288serves as the resource domain for the entire subtree. 289 290Inside a threaded subtree, threads of a process can be put in 291different cgroups and are not subject to the no internal process 292constraint - threaded controllers can be enabled on non-leaf cgroups 293whether they have threads in them or not. 294 295As the threaded domain cgroup hosts all the domain resource 296consumptions of the subtree, it is considered to have internal 297resource consumptions whether there are processes in it or not and 298can't have populated child cgroups which aren't threaded. Because the 299root cgroup is not subject to no internal process constraint, it can 300serve both as a threaded domain and a parent to domain cgroups. 301 302The current operation mode or type of the cgroup is shown in the 303"cgroup.type" file which indicates whether the cgroup is a normal 304domain, a domain which is serving as the domain of a threaded subtree, 305or a threaded cgroup. 306 307On creation, a cgroup is always a domain cgroup and can be made 308threaded by writing "threaded" to the "cgroup.type" file. The 309operation is single direction:: 310 311 # echo threaded > cgroup.type 312 313Once threaded, the cgroup can't be made a domain again. To enable the 314thread mode, the following conditions must be met. 315 316- As the cgroup will join the parent's resource domain. The parent 317 must either be a valid (threaded) domain or a threaded cgroup. 318 319- When the parent is an unthreaded domain, it must not have any domain 320 controllers enabled or populated domain children. The root is 321 exempt from this requirement. 322 323Topology-wise, a cgroup can be in an invalid state. Please consider 324the following topology:: 325 326 A (threaded domain) - B (threaded) - C (domain, just created) 327 328C is created as a domain but isn't connected to a parent which can 329host child domains. C can't be used until it is turned into a 330threaded cgroup. "cgroup.type" file will report "domain (invalid)" in 331these cases. Operations which fail due to invalid topology use 332EOPNOTSUPP as the errno. 333 334A domain cgroup is turned into a threaded domain when one of its child 335cgroup becomes threaded or threaded controllers are enabled in the 336"cgroup.subtree_control" file while there are processes in the cgroup. 337A threaded domain reverts to a normal domain when the conditions 338clear. 339 340When read, "cgroup.threads" contains the list of the thread IDs of all 341threads in the cgroup. Except that the operations are per-thread 342instead of per-process, "cgroup.threads" has the same format and 343behaves the same way as "cgroup.procs". While "cgroup.threads" can be 344written to in any cgroup, as it can only move threads inside the same 345threaded domain, its operations are confined inside each threaded 346subtree. 347 348The threaded domain cgroup serves as the resource domain for the whole 349subtree, and, while the threads can be scattered across the subtree, 350all the processes are considered to be in the threaded domain cgroup. 351"cgroup.procs" in a threaded domain cgroup contains the PIDs of all 352processes in the subtree and is not readable in the subtree proper. 353However, "cgroup.procs" can be written to from anywhere in the subtree 354to migrate all threads of the matching process to the cgroup. 355 356Only threaded controllers can be enabled in a threaded subtree. When 357a threaded controller is enabled inside a threaded subtree, it only 358accounts for and controls resource consumptions associated with the 359threads in the cgroup and its descendants. All consumptions which 360aren't tied to a specific thread belong to the threaded domain cgroup. 361 362Because a threaded subtree is exempt from no internal process 363constraint, a threaded controller must be able to handle competition 364between threads in a non-leaf cgroup and its child cgroups. Each 365threaded controller defines how such competitions are handled. 366 367 368[Un]populated Notification 369-------------------------- 370 371Each non-root cgroup has a "cgroup.events" file which contains 372"populated" field indicating whether the cgroup's sub-hierarchy has 373live processes in it. Its value is 0 if there is no live process in 374the cgroup and its descendants; otherwise, 1. poll and [id]notify 375events are triggered when the value changes. This can be used, for 376example, to start a clean-up operation after all processes of a given 377sub-hierarchy have exited. The populated state updates and 378notifications are recursive. Consider the following sub-hierarchy 379where the numbers in the parentheses represent the numbers of processes 380in each cgroup:: 381 382 A(4) - B(0) - C(1) 383 \ D(0) 384 385A, B and C's "populated" fields would be 1 while D's 0. After the one 386process in C exits, B and C's "populated" fields would flip to "0" and 387file modified events will be generated on the "cgroup.events" files of 388both cgroups. 389 390 391Controlling Controllers 392----------------------- 393 394Enabling and Disabling 395~~~~~~~~~~~~~~~~~~~~~~ 396 397Each cgroup has a "cgroup.controllers" file which lists all 398controllers available for the cgroup to enable:: 399 400 # cat cgroup.controllers 401 cpu io memory 402 403No controller is enabled by default. Controllers can be enabled and 404disabled by writing to the "cgroup.subtree_control" file:: 405 406 # echo "+cpu +memory -io" > cgroup.subtree_control 407 408Only controllers which are listed in "cgroup.controllers" can be 409enabled. When multiple operations are specified as above, either they 410all succeed or fail. If multiple operations on the same controller 411are specified, the last one is effective. 412 413Enabling a controller in a cgroup indicates that the distribution of 414the target resource across its immediate children will be controlled. 415Consider the following sub-hierarchy. The enabled controllers are 416listed in parentheses:: 417 418 A(cpu,memory) - B(memory) - C() 419 \ D() 420 421As A has "cpu" and "memory" enabled, A will control the distribution 422of CPU cycles and memory to its children, in this case, B. As B has 423"memory" enabled but not "CPU", C and D will compete freely on CPU 424cycles but their division of memory available to B will be controlled. 425 426As a controller regulates the distribution of the target resource to 427the cgroup's children, enabling it creates the controller's interface 428files in the child cgroups. In the above example, enabling "cpu" on B 429would create the "cpu." prefixed controller interface files in C and 430D. Likewise, disabling "memory" from B would remove the "memory." 431prefixed controller interface files from C and D. This means that the 432controller interface files - anything which doesn't start with 433"cgroup." are owned by the parent rather than the cgroup itself. 434 435 436Top-down Constraint 437~~~~~~~~~~~~~~~~~~~ 438 439Resources are distributed top-down and a cgroup can further distribute 440a resource only if the resource has been distributed to it from the 441parent. This means that all non-root "cgroup.subtree_control" files 442can only contain controllers which are enabled in the parent's 443"cgroup.subtree_control" file. A controller can be enabled only if 444the parent has the controller enabled and a controller can't be 445disabled if one or more children have it enabled. 446 447 448No Internal Process Constraint 449~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 450 451Non-root cgroups can distribute domain resources to their children 452only when they don't have any processes of their own. In other words, 453only domain cgroups which don't contain any processes can have domain 454controllers enabled in their "cgroup.subtree_control" files. 455 456This guarantees that, when a domain controller is looking at the part 457of the hierarchy which has it enabled, processes are always only on 458the leaves. This rules out situations where child cgroups compete 459against internal processes of the parent. 460 461The root cgroup is exempt from this restriction. Root contains 462processes and anonymous resource consumption which can't be associated 463with any other cgroups and requires special treatment from most 464controllers. How resource consumption in the root cgroup is governed 465is up to each controller (for more information on this topic please 466refer to the Non-normative information section in the Controllers 467chapter). 468 469Note that the restriction doesn't get in the way if there is no 470enabled controller in the cgroup's "cgroup.subtree_control". This is 471important as otherwise it wouldn't be possible to create children of a 472populated cgroup. To control resource distribution of a cgroup, the 473cgroup must create children and transfer all its processes to the 474children before enabling controllers in its "cgroup.subtree_control" 475file. 476 477 478Delegation 479---------- 480 481Model of Delegation 482~~~~~~~~~~~~~~~~~~~ 483 484A cgroup can be delegated in two ways. First, to a less privileged 485user by granting write access of the directory and its "cgroup.procs", 486"cgroup.threads" and "cgroup.subtree_control" files to the user. 487Second, if the "nsdelegate" mount option is set, automatically to a 488cgroup namespace on namespace creation. 489 490Because the resource control interface files in a given directory 491control the distribution of the parent's resources, the delegatee 492shouldn't be allowed to write to them. For the first method, this is 493achieved by not granting access to these files. For the second, the 494kernel rejects writes to all files other than "cgroup.procs" and 495"cgroup.subtree_control" on a namespace root from inside the 496namespace. 497 498The end results are equivalent for both delegation types. Once 499delegated, the user can build sub-hierarchy under the directory, 500organize processes inside it as it sees fit and further distribute the 501resources it received from the parent. The limits and other settings 502of all resource controllers are hierarchical and regardless of what 503happens in the delegated sub-hierarchy, nothing can escape the 504resource restrictions imposed by the parent. 505 506Currently, cgroup doesn't impose any restrictions on the number of 507cgroups in or nesting depth of a delegated sub-hierarchy; however, 508this may be limited explicitly in the future. 509 510 511Delegation Containment 512~~~~~~~~~~~~~~~~~~~~~~ 513 514A delegated sub-hierarchy is contained in the sense that processes 515can't be moved into or out of the sub-hierarchy by the delegatee. 516 517For delegations to a less privileged user, this is achieved by 518requiring the following conditions for a process with a non-root euid 519to migrate a target process into a cgroup by writing its PID to the 520"cgroup.procs" file. 521 522- The writer must have write access to the "cgroup.procs" file. 523 524- The writer must have write access to the "cgroup.procs" file of the 525 common ancestor of the source and destination cgroups. 526 527The above two constraints ensure that while a delegatee may migrate 528processes around freely in the delegated sub-hierarchy it can't pull 529in from or push out to outside the sub-hierarchy. 530 531For an example, let's assume cgroups C0 and C1 have been delegated to 532user U0 who created C00, C01 under C0 and C10 under C1 as follows and 533all processes under C0 and C1 belong to U0:: 534 535 ~~~~~~~~~~~~~ - C0 - C00 536 ~ cgroup ~ \ C01 537 ~ hierarchy ~ 538 ~~~~~~~~~~~~~ - C1 - C10 539 540Let's also say U0 wants to write the PID of a process which is 541currently in C10 into "C00/cgroup.procs". U0 has write access to the 542file; however, the common ancestor of the source cgroup C10 and the 543destination cgroup C00 is above the points of delegation and U0 would 544not have write access to its "cgroup.procs" files and thus the write 545will be denied with -EACCES. 546 547For delegations to namespaces, containment is achieved by requiring 548that both the source and destination cgroups are reachable from the 549namespace of the process which is attempting the migration. If either 550is not reachable, the migration is rejected with -ENOENT. 551 552 553Guidelines 554---------- 555 556Organize Once and Control 557~~~~~~~~~~~~~~~~~~~~~~~~~ 558 559Migrating a process across cgroups is a relatively expensive operation 560and stateful resources such as memory are not moved together with the 561process. This is an explicit design decision as there often exist 562inherent trade-offs between migration and various hot paths in terms 563of synchronization cost. 564 565As such, migrating processes across cgroups frequently as a means to 566apply different resource restrictions is discouraged. A workload 567should be assigned to a cgroup according to the system's logical and 568resource structure once on start-up. Dynamic adjustments to resource 569distribution can be made by changing controller configuration through 570the interface files. 571 572 573Avoid Name Collisions 574~~~~~~~~~~~~~~~~~~~~~ 575 576Interface files for a cgroup and its children cgroups occupy the same 577directory and it is possible to create children cgroups which collide 578with interface files. 579 580All cgroup core interface files are prefixed with "cgroup." and each 581controller's interface files are prefixed with the controller name and 582a dot. A controller's name is composed of lower case alphabets and 583'_'s but never begins with an '_' so it can be used as the prefix 584character for collision avoidance. Also, interface file names won't 585start or end with terms which are often used in categorizing workloads 586such as job, service, slice, unit or workload. 587 588cgroup doesn't do anything to prevent name collisions and it's the 589user's responsibility to avoid them. 590 591 592Resource Distribution Models 593============================ 594 595cgroup controllers implement several resource distribution schemes 596depending on the resource type and expected use cases. This section 597describes major schemes in use along with their expected behaviors. 598 599 600Weights 601------- 602 603A parent's resource is distributed by adding up the weights of all 604active children and giving each the fraction matching the ratio of its 605weight against the sum. As only children which can make use of the 606resource at the moment participate in the distribution, this is 607work-conserving. Due to the dynamic nature, this model is usually 608used for stateless resources. 609 610All weights are in the range [1, 10000] with the default at 100. This 611allows symmetric multiplicative biases in both directions at fine 612enough granularity while staying in the intuitive range. 613 614As long as the weight is in range, all configuration combinations are 615valid and there is no reason to reject configuration changes or 616process migrations. 617 618"cpu.weight" proportionally distributes CPU cycles to active children 619and is an example of this type. 620 621 622.. _cgroupv2-limits-distributor: 623 624Limits 625------ 626 627A child can only consume upto the configured amount of the resource. 628Limits can be over-committed - the sum of the limits of children can 629exceed the amount of resource available to the parent. 630 631Limits are in the range [0, max] and defaults to "max", which is noop. 632 633As limits can be over-committed, all configuration combinations are 634valid and there is no reason to reject configuration changes or 635process migrations. 636 637"io.max" limits the maximum BPS and/or IOPS that a cgroup can consume 638on an IO device and is an example of this type. 639 640.. _cgroupv2-protections-distributor: 641 642Protections 643----------- 644 645A cgroup is protected upto the configured amount of the resource 646as long as the usages of all its ancestors are under their 647protected levels. Protections can be hard guarantees or best effort 648soft boundaries. Protections can also be over-committed in which case 649only upto the amount available to the parent is protected among 650children. 651 652Protections are in the range [0, max] and defaults to 0, which is 653noop. 654 655As protections can be over-committed, all configuration combinations 656are valid and there is no reason to reject configuration changes or 657process migrations. 658 659"memory.low" implements best-effort memory protection and is an 660example of this type. 661 662 663Allocations 664----------- 665 666A cgroup is exclusively allocated a certain amount of a finite 667resource. Allocations can't be over-committed - the sum of the 668allocations of children can not exceed the amount of resource 669available to the parent. 670 671Allocations are in the range [0, max] and defaults to 0, which is no 672resource. 673 674As allocations can't be over-committed, some configuration 675combinations are invalid and should be rejected. Also, if the 676resource is mandatory for execution of processes, process migrations 677may be rejected. 678 679"cpu.rt.max" hard-allocates realtime slices and is an example of this 680type. 681 682 683Interface Files 684=============== 685 686Format 687------ 688 689All interface files should be in one of the following formats whenever 690possible:: 691 692 New-line separated values 693 (when only one value can be written at once) 694 695 VAL0\n 696 VAL1\n 697 ... 698 699 Space separated values 700 (when read-only or multiple values can be written at once) 701 702 VAL0 VAL1 ...\n 703 704 Flat keyed 705 706 KEY0 VAL0\n 707 KEY1 VAL1\n 708 ... 709 710 Nested keyed 711 712 KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01... 713 KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11... 714 ... 715 716For a writable file, the format for writing should generally match 717reading; however, controllers may allow omitting later fields or 718implement restricted shortcuts for most common use cases. 719 720For both flat and nested keyed files, only the values for a single key 721can be written at a time. For nested keyed files, the sub key pairs 722may be specified in any order and not all pairs have to be specified. 723 724 725Conventions 726----------- 727 728- Settings for a single feature should be contained in a single file. 729 730- The root cgroup should be exempt from resource control and thus 731 shouldn't have resource control interface files. 732 733- The default time unit is microseconds. If a different unit is ever 734 used, an explicit unit suffix must be present. 735 736- A parts-per quantity should use a percentage decimal with at least 737 two digit fractional part - e.g. 13.40. 738 739- If a controller implements weight based resource distribution, its 740 interface file should be named "weight" and have the range [1, 741 10000] with 100 as the default. The values are chosen to allow 742 enough and symmetric bias in both directions while keeping it 743 intuitive (the default is 100%). 744 745- If a controller implements an absolute resource guarantee and/or 746 limit, the interface files should be named "min" and "max" 747 respectively. If a controller implements best effort resource 748 guarantee and/or limit, the interface files should be named "low" 749 and "high" respectively. 750 751 In the above four control files, the special token "max" should be 752 used to represent upward infinity for both reading and writing. 753 754- If a setting has a configurable default value and keyed specific 755 overrides, the default entry should be keyed with "default" and 756 appear as the first entry in the file. 757 758 The default value can be updated by writing either "default $VAL" or 759 "$VAL". 760 761 When writing to update a specific override, "default" can be used as 762 the value to indicate removal of the override. Override entries 763 with "default" as the value must not appear when read. 764 765 For example, a setting which is keyed by major:minor device numbers 766 with integer values may look like the following:: 767 768 # cat cgroup-example-interface-file 769 default 150 770 8:0 300 771 772 The default value can be updated by:: 773 774 # echo 125 > cgroup-example-interface-file 775 776 or:: 777 778 # echo "default 125" > cgroup-example-interface-file 779 780 An override can be set by:: 781 782 # echo "8:16 170" > cgroup-example-interface-file 783 784 and cleared by:: 785 786 # echo "8:0 default" > cgroup-example-interface-file 787 # cat cgroup-example-interface-file 788 default 125 789 8:16 170 790 791- For events which are not very high frequency, an interface file 792 "events" should be created which lists event key value pairs. 793 Whenever a notifiable event happens, file modified event should be 794 generated on the file. 795 796 797Core Interface Files 798-------------------- 799 800All cgroup core files are prefixed with "cgroup." 801 802 cgroup.type 803 A read-write single value file which exists on non-root 804 cgroups. 805 806 When read, it indicates the current type of the cgroup, which 807 can be one of the following values. 808 809 - "domain" : A normal valid domain cgroup. 810 811 - "domain threaded" : A threaded domain cgroup which is 812 serving as the root of a threaded subtree. 813 814 - "domain invalid" : A cgroup which is in an invalid state. 815 It can't be populated or have controllers enabled. It may 816 be allowed to become a threaded cgroup. 817 818 - "threaded" : A threaded cgroup which is a member of a 819 threaded subtree. 820 821 A cgroup can be turned into a threaded cgroup by writing 822 "threaded" to this file. 823 824 cgroup.procs 825 A read-write new-line separated values file which exists on 826 all cgroups. 827 828 When read, it lists the PIDs of all processes which belong to 829 the cgroup one-per-line. The PIDs are not ordered and the 830 same PID may show up more than once if the process got moved 831 to another cgroup and then back or the PID got recycled while 832 reading. 833 834 A PID can be written to migrate the process associated with 835 the PID to the cgroup. The writer should match all of the 836 following conditions. 837 838 - It must have write access to the "cgroup.procs" file. 839 840 - It must have write access to the "cgroup.procs" file of the 841 common ancestor of the source and destination cgroups. 842 843 When delegating a sub-hierarchy, write access to this file 844 should be granted along with the containing directory. 845 846 In a threaded cgroup, reading this file fails with EOPNOTSUPP 847 as all the processes belong to the thread root. Writing is 848 supported and moves every thread of the process to the cgroup. 849 850 cgroup.threads 851 A read-write new-line separated values file which exists on 852 all cgroups. 853 854 When read, it lists the TIDs of all threads which belong to 855 the cgroup one-per-line. The TIDs are not ordered and the 856 same TID may show up more than once if the thread got moved to 857 another cgroup and then back or the TID got recycled while 858 reading. 859 860 A TID can be written to migrate the thread associated with the 861 TID to the cgroup. The writer should match all of the 862 following conditions. 863 864 - It must have write access to the "cgroup.threads" file. 865 866 - The cgroup that the thread is currently in must be in the 867 same resource domain as the destination cgroup. 868 869 - It must have write access to the "cgroup.procs" file of the 870 common ancestor of the source and destination cgroups. 871 872 When delegating a sub-hierarchy, write access to this file 873 should be granted along with the containing directory. 874 875 cgroup.controllers 876 A read-only space separated values file which exists on all 877 cgroups. 878 879 It shows space separated list of all controllers available to 880 the cgroup. The controllers are not ordered. 881 882 cgroup.subtree_control 883 A read-write space separated values file which exists on all 884 cgroups. Starts out empty. 885 886 When read, it shows space separated list of the controllers 887 which are enabled to control resource distribution from the 888 cgroup to its children. 889 890 Space separated list of controllers prefixed with '+' or '-' 891 can be written to enable or disable controllers. A controller 892 name prefixed with '+' enables the controller and '-' 893 disables. If a controller appears more than once on the list, 894 the last one is effective. When multiple enable and disable 895 operations are specified, either all succeed or all fail. 896 897 cgroup.events 898 A read-only flat-keyed file which exists on non-root cgroups. 899 The following entries are defined. Unless specified 900 otherwise, a value change in this file generates a file 901 modified event. 902 903 populated 904 1 if the cgroup or its descendants contains any live 905 processes; otherwise, 0. 906 frozen 907 1 if the cgroup is frozen; otherwise, 0. 908 909 cgroup.max.descendants 910 A read-write single value files. The default is "max". 911 912 Maximum allowed number of descent cgroups. 913 If the actual number of descendants is equal or larger, 914 an attempt to create a new cgroup in the hierarchy will fail. 915 916 cgroup.max.depth 917 A read-write single value files. The default is "max". 918 919 Maximum allowed descent depth below the current cgroup. 920 If the actual descent depth is equal or larger, 921 an attempt to create a new child cgroup will fail. 922 923 cgroup.stat 924 A read-only flat-keyed file with the following entries: 925 926 nr_descendants 927 Total number of visible descendant cgroups. 928 929 nr_dying_descendants 930 Total number of dying descendant cgroups. A cgroup becomes 931 dying after being deleted by a user. The cgroup will remain 932 in dying state for some time undefined time (which can depend 933 on system load) before being completely destroyed. 934 935 A process can't enter a dying cgroup under any circumstances, 936 a dying cgroup can't revive. 937 938 A dying cgroup can consume system resources not exceeding 939 limits, which were active at the moment of cgroup deletion. 940 941 cgroup.freeze 942 A read-write single value file which exists on non-root cgroups. 943 Allowed values are "0" and "1". The default is "0". 944 945 Writing "1" to the file causes freezing of the cgroup and all 946 descendant cgroups. This means that all belonging processes will 947 be stopped and will not run until the cgroup will be explicitly 948 unfrozen. Freezing of the cgroup may take some time; when this action 949 is completed, the "frozen" value in the cgroup.events control file 950 will be updated to "1" and the corresponding notification will be 951 issued. 952 953 A cgroup can be frozen either by its own settings, or by settings 954 of any ancestor cgroups. If any of ancestor cgroups is frozen, the 955 cgroup will remain frozen. 956 957 Processes in the frozen cgroup can be killed by a fatal signal. 958 They also can enter and leave a frozen cgroup: either by an explicit 959 move by a user, or if freezing of the cgroup races with fork(). 960 If a process is moved to a frozen cgroup, it stops. If a process is 961 moved out of a frozen cgroup, it becomes running. 962 963 Frozen status of a cgroup doesn't affect any cgroup tree operations: 964 it's possible to delete a frozen (and empty) cgroup, as well as 965 create new sub-cgroups. 966 967 cgroup.kill 968 A write-only single value file which exists in non-root cgroups. 969 The only allowed value is "1". 970 971 Writing "1" to the file causes the cgroup and all descendant cgroups to 972 be killed. This means that all processes located in the affected cgroup 973 tree will be killed via SIGKILL. 974 975 Killing a cgroup tree will deal with concurrent forks appropriately and 976 is protected against migrations. 977 978 In a threaded cgroup, writing this file fails with EOPNOTSUPP as 979 killing cgroups is a process directed operation, i.e. it affects 980 the whole thread-group. 981 982 cgroup.pressure 983 A read-write single value file that allowed values are "0" and "1". 984 The default is "1". 985 986 Writing "0" to the file will disable the cgroup PSI accounting. 987 Writing "1" to the file will re-enable the cgroup PSI accounting. 988 989 This control attribute is not hierarchical, so disable or enable PSI 990 accounting in a cgroup does not affect PSI accounting in descendants 991 and doesn't need pass enablement via ancestors from root. 992 993 The reason this control attribute exists is that PSI accounts stalls for 994 each cgroup separately and aggregates it at each level of the hierarchy. 995 This may cause non-negligible overhead for some workloads when under 996 deep level of the hierarchy, in which case this control attribute can 997 be used to disable PSI accounting in the non-leaf cgroups. 998 999 irq.pressure 1000 A read-write nested-keyed file. 1001 1002 Shows pressure stall information for IRQ/SOFTIRQ. See 1003 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1004 1005Controllers 1006=========== 1007 1008.. _cgroup-v2-cpu: 1009 1010CPU 1011--- 1012 1013The "cpu" controllers regulates distribution of CPU cycles. This 1014controller implements weight and absolute bandwidth limit models for 1015normal scheduling policy and absolute bandwidth allocation model for 1016realtime scheduling policy. 1017 1018In all the above models, cycles distribution is defined only on a temporal 1019base and it does not account for the frequency at which tasks are executed. 1020The (optional) utilization clamping support allows to hint the schedutil 1021cpufreq governor about the minimum desired frequency which should always be 1022provided by a CPU, as well as the maximum desired frequency, which should not 1023be exceeded by a CPU. 1024 1025WARNING: cgroup2 doesn't yet support control of realtime processes and 1026the cpu controller can only be enabled when all RT processes are in 1027the root cgroup. Be aware that system management software may already 1028have placed RT processes into nonroot cgroups during the system boot 1029process, and these processes may need to be moved to the root cgroup 1030before the cpu controller can be enabled. 1031 1032 1033CPU Interface Files 1034~~~~~~~~~~~~~~~~~~~ 1035 1036All time durations are in microseconds. 1037 1038 cpu.stat 1039 A read-only flat-keyed file. 1040 This file exists whether the controller is enabled or not. 1041 1042 It always reports the following three stats: 1043 1044 - usage_usec 1045 - user_usec 1046 - system_usec 1047 1048 and the following three when the controller is enabled: 1049 1050 - nr_periods 1051 - nr_throttled 1052 - throttled_usec 1053 - nr_bursts 1054 - burst_usec 1055 1056 cpu.weight 1057 A read-write single value file which exists on non-root 1058 cgroups. The default is "100". 1059 1060 The weight in the range [1, 10000]. 1061 1062 cpu.weight.nice 1063 A read-write single value file which exists on non-root 1064 cgroups. The default is "0". 1065 1066 The nice value is in the range [-20, 19]. 1067 1068 This interface file is an alternative interface for 1069 "cpu.weight" and allows reading and setting weight using the 1070 same values used by nice(2). Because the range is smaller and 1071 granularity is coarser for the nice values, the read value is 1072 the closest approximation of the current weight. 1073 1074 cpu.max 1075 A read-write two value file which exists on non-root cgroups. 1076 The default is "max 100000". 1077 1078 The maximum bandwidth limit. It's in the following format:: 1079 1080 $MAX $PERIOD 1081 1082 which indicates that the group may consume upto $MAX in each 1083 $PERIOD duration. "max" for $MAX indicates no limit. If only 1084 one number is written, $MAX is updated. 1085 1086 cpu.max.burst 1087 A read-write single value file which exists on non-root 1088 cgroups. The default is "0". 1089 1090 The burst in the range [0, $MAX]. 1091 1092 cpu.pressure 1093 A read-write nested-keyed file. 1094 1095 Shows pressure stall information for CPU. See 1096 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1097 1098 cpu.uclamp.min 1099 A read-write single value file which exists on non-root cgroups. 1100 The default is "0", i.e. no utilization boosting. 1101 1102 The requested minimum utilization (protection) as a percentage 1103 rational number, e.g. 12.34 for 12.34%. 1104 1105 This interface allows reading and setting minimum utilization clamp 1106 values similar to the sched_setattr(2). This minimum utilization 1107 value is used to clamp the task specific minimum utilization clamp. 1108 1109 The requested minimum utilization (protection) is always capped by 1110 the current value for the maximum utilization (limit), i.e. 1111 `cpu.uclamp.max`. 1112 1113 cpu.uclamp.max 1114 A read-write single value file which exists on non-root cgroups. 1115 The default is "max". i.e. no utilization capping 1116 1117 The requested maximum utilization (limit) as a percentage rational 1118 number, e.g. 98.76 for 98.76%. 1119 1120 This interface allows reading and setting maximum utilization clamp 1121 values similar to the sched_setattr(2). This maximum utilization 1122 value is used to clamp the task specific maximum utilization clamp. 1123 1124 1125 1126Memory 1127------ 1128 1129The "memory" controller regulates distribution of memory. Memory is 1130stateful and implements both limit and protection models. Due to the 1131intertwining between memory usage and reclaim pressure and the 1132stateful nature of memory, the distribution model is relatively 1133complex. 1134 1135While not completely water-tight, all major memory usages by a given 1136cgroup are tracked so that the total memory consumption can be 1137accounted and controlled to a reasonable extent. Currently, the 1138following types of memory usages are tracked. 1139 1140- Userland memory - page cache and anonymous memory. 1141 1142- Kernel data structures such as dentries and inodes. 1143 1144- TCP socket buffers. 1145 1146The above list may expand in the future for better coverage. 1147 1148 1149Memory Interface Files 1150~~~~~~~~~~~~~~~~~~~~~~ 1151 1152All memory amounts are in bytes. If a value which is not aligned to 1153PAGE_SIZE is written, the value may be rounded up to the closest 1154PAGE_SIZE multiple when read back. 1155 1156 memory.current 1157 A read-only single value file which exists on non-root 1158 cgroups. 1159 1160 The total amount of memory currently being used by the cgroup 1161 and its descendants. 1162 1163 memory.min 1164 A read-write single value file which exists on non-root 1165 cgroups. The default is "0". 1166 1167 Hard memory protection. If the memory usage of a cgroup 1168 is within its effective min boundary, the cgroup's memory 1169 won't be reclaimed under any conditions. If there is no 1170 unprotected reclaimable memory available, OOM killer 1171 is invoked. Above the effective min boundary (or 1172 effective low boundary if it is higher), pages are reclaimed 1173 proportionally to the overage, reducing reclaim pressure for 1174 smaller overages. 1175 1176 Effective min boundary is limited by memory.min values of 1177 all ancestor cgroups. If there is memory.min overcommitment 1178 (child cgroup or cgroups are requiring more protected memory 1179 than parent will allow), then each child cgroup will get 1180 the part of parent's protection proportional to its 1181 actual memory usage below memory.min. 1182 1183 Putting more memory than generally available under this 1184 protection is discouraged and may lead to constant OOMs. 1185 1186 If a memory cgroup is not populated with processes, 1187 its memory.min is ignored. 1188 1189 memory.low 1190 A read-write single value file which exists on non-root 1191 cgroups. The default is "0". 1192 1193 Best-effort memory protection. If the memory usage of a 1194 cgroup is within its effective low boundary, the cgroup's 1195 memory won't be reclaimed unless there is no reclaimable 1196 memory available in unprotected cgroups. 1197 Above the effective low boundary (or 1198 effective min boundary if it is higher), pages are reclaimed 1199 proportionally to the overage, reducing reclaim pressure for 1200 smaller overages. 1201 1202 Effective low boundary is limited by memory.low values of 1203 all ancestor cgroups. If there is memory.low overcommitment 1204 (child cgroup or cgroups are requiring more protected memory 1205 than parent will allow), then each child cgroup will get 1206 the part of parent's protection proportional to its 1207 actual memory usage below memory.low. 1208 1209 Putting more memory than generally available under this 1210 protection is discouraged. 1211 1212 memory.high 1213 A read-write single value file which exists on non-root 1214 cgroups. The default is "max". 1215 1216 Memory usage throttle limit. This is the main mechanism to 1217 control memory usage of a cgroup. If a cgroup's usage goes 1218 over the high boundary, the processes of the cgroup are 1219 throttled and put under heavy reclaim pressure. 1220 1221 Going over the high limit never invokes the OOM killer and 1222 under extreme conditions the limit may be breached. 1223 1224 memory.max 1225 A read-write single value file which exists on non-root 1226 cgroups. The default is "max". 1227 1228 Memory usage hard limit. This is the final protection 1229 mechanism. If a cgroup's memory usage reaches this limit and 1230 can't be reduced, the OOM killer is invoked in the cgroup. 1231 Under certain circumstances, the usage may go over the limit 1232 temporarily. 1233 1234 In default configuration regular 0-order allocations always 1235 succeed unless OOM killer chooses current task as a victim. 1236 1237 Some kinds of allocations don't invoke the OOM killer. 1238 Caller could retry them differently, return into userspace 1239 as -ENOMEM or silently ignore in cases like disk readahead. 1240 1241 This is the ultimate protection mechanism. As long as the 1242 high limit is used and monitored properly, this limit's 1243 utility is limited to providing the final safety net. 1244 1245 memory.reclaim 1246 A write-only nested-keyed file which exists for all cgroups. 1247 1248 This is a simple interface to trigger memory reclaim in the 1249 target cgroup. 1250 1251 This file accepts a string which contains the number of bytes to 1252 reclaim. 1253 1254 Example:: 1255 1256 echo "1G" > memory.reclaim 1257 1258 Please note that the kernel can over or under reclaim from 1259 the target cgroup. If less bytes are reclaimed than the 1260 specified amount, -EAGAIN is returned. 1261 1262 Please note that the proactive reclaim (triggered by this 1263 interface) is not meant to indicate memory pressure on the 1264 memory cgroup. Therefore socket memory balancing triggered by 1265 the memory reclaim normally is not exercised in this case. 1266 This means that the networking layer will not adapt based on 1267 reclaim induced by memory.reclaim. 1268 1269 This file also allows the user to specify the nodes to reclaim from, 1270 via the 'nodes=' key, for example:: 1271 1272 echo "1G nodes=0,1" > memory.reclaim 1273 1274 The above instructs the kernel to reclaim memory from nodes 0,1. 1275 1276 memory.peak 1277 A read-only single value file which exists on non-root 1278 cgroups. 1279 1280 The max memory usage recorded for the cgroup and its 1281 descendants since the creation of the cgroup. 1282 1283 memory.oom.group 1284 A read-write single value file which exists on non-root 1285 cgroups. The default value is "0". 1286 1287 Determines whether the cgroup should be treated as 1288 an indivisible workload by the OOM killer. If set, 1289 all tasks belonging to the cgroup or to its descendants 1290 (if the memory cgroup is not a leaf cgroup) are killed 1291 together or not at all. This can be used to avoid 1292 partial kills to guarantee workload integrity. 1293 1294 Tasks with the OOM protection (oom_score_adj set to -1000) 1295 are treated as an exception and are never killed. 1296 1297 If the OOM killer is invoked in a cgroup, it's not going 1298 to kill any tasks outside of this cgroup, regardless 1299 memory.oom.group values of ancestor cgroups. 1300 1301 memory.events 1302 A read-only flat-keyed file which exists on non-root cgroups. 1303 The following entries are defined. Unless specified 1304 otherwise, a value change in this file generates a file 1305 modified event. 1306 1307 Note that all fields in this file are hierarchical and the 1308 file modified event can be generated due to an event down the 1309 hierarchy. For the local events at the cgroup level see 1310 memory.events.local. 1311 1312 low 1313 The number of times the cgroup is reclaimed due to 1314 high memory pressure even though its usage is under 1315 the low boundary. This usually indicates that the low 1316 boundary is over-committed. 1317 1318 high 1319 The number of times processes of the cgroup are 1320 throttled and routed to perform direct memory reclaim 1321 because the high memory boundary was exceeded. For a 1322 cgroup whose memory usage is capped by the high limit 1323 rather than global memory pressure, this event's 1324 occurrences are expected. 1325 1326 max 1327 The number of times the cgroup's memory usage was 1328 about to go over the max boundary. If direct reclaim 1329 fails to bring it down, the cgroup goes to OOM state. 1330 1331 oom 1332 The number of time the cgroup's memory usage was 1333 reached the limit and allocation was about to fail. 1334 1335 This event is not raised if the OOM killer is not 1336 considered as an option, e.g. for failed high-order 1337 allocations or if caller asked to not retry attempts. 1338 1339 oom_kill 1340 The number of processes belonging to this cgroup 1341 killed by any kind of OOM killer. 1342 1343 oom_group_kill 1344 The number of times a group OOM has occurred. 1345 1346 memory.events.local 1347 Similar to memory.events but the fields in the file are local 1348 to the cgroup i.e. not hierarchical. The file modified event 1349 generated on this file reflects only the local events. 1350 1351 memory.stat 1352 A read-only flat-keyed file which exists on non-root cgroups. 1353 1354 This breaks down the cgroup's memory footprint into different 1355 types of memory, type-specific details, and other information 1356 on the state and past events of the memory management system. 1357 1358 All memory amounts are in bytes. 1359 1360 The entries are ordered to be human readable, and new entries 1361 can show up in the middle. Don't rely on items remaining in a 1362 fixed position; use the keys to look up specific values! 1363 1364 If the entry has no per-node counter (or not show in the 1365 memory.numa_stat). We use 'npn' (non-per-node) as the tag 1366 to indicate that it will not show in the memory.numa_stat. 1367 1368 anon 1369 Amount of memory used in anonymous mappings such as 1370 brk(), sbrk(), and mmap(MAP_ANONYMOUS) 1371 1372 file 1373 Amount of memory used to cache filesystem data, 1374 including tmpfs and shared memory. 1375 1376 kernel (npn) 1377 Amount of total kernel memory, including 1378 (kernel_stack, pagetables, percpu, vmalloc, slab) in 1379 addition to other kernel memory use cases. 1380 1381 kernel_stack 1382 Amount of memory allocated to kernel stacks. 1383 1384 pagetables 1385 Amount of memory allocated for page tables. 1386 1387 sec_pagetables 1388 Amount of memory allocated for secondary page tables, 1389 this currently includes KVM mmu allocations on x86 1390 and arm64. 1391 1392 percpu (npn) 1393 Amount of memory used for storing per-cpu kernel 1394 data structures. 1395 1396 sock (npn) 1397 Amount of memory used in network transmission buffers 1398 1399 vmalloc (npn) 1400 Amount of memory used for vmap backed memory. 1401 1402 shmem 1403 Amount of cached filesystem data that is swap-backed, 1404 such as tmpfs, shm segments, shared anonymous mmap()s 1405 1406 zswap 1407 Amount of memory consumed by the zswap compression backend. 1408 1409 zswapped 1410 Amount of application memory swapped out to zswap. 1411 1412 file_mapped 1413 Amount of cached filesystem data mapped with mmap() 1414 1415 file_dirty 1416 Amount of cached filesystem data that was modified but 1417 not yet written back to disk 1418 1419 file_writeback 1420 Amount of cached filesystem data that was modified and 1421 is currently being written back to disk 1422 1423 swapcached 1424 Amount of swap cached in memory. The swapcache is accounted 1425 against both memory and swap usage. 1426 1427 anon_thp 1428 Amount of memory used in anonymous mappings backed by 1429 transparent hugepages 1430 1431 file_thp 1432 Amount of cached filesystem data backed by transparent 1433 hugepages 1434 1435 shmem_thp 1436 Amount of shm, tmpfs, shared anonymous mmap()s backed by 1437 transparent hugepages 1438 1439 inactive_anon, active_anon, inactive_file, active_file, unevictable 1440 Amount of memory, swap-backed and filesystem-backed, 1441 on the internal memory management lists used by the 1442 page reclaim algorithm. 1443 1444 As these represent internal list state (eg. shmem pages are on anon 1445 memory management lists), inactive_foo + active_foo may not be equal to 1446 the value for the foo counter, since the foo counter is type-based, not 1447 list-based. 1448 1449 slab_reclaimable 1450 Part of "slab" that might be reclaimed, such as 1451 dentries and inodes. 1452 1453 slab_unreclaimable 1454 Part of "slab" that cannot be reclaimed on memory 1455 pressure. 1456 1457 slab (npn) 1458 Amount of memory used for storing in-kernel data 1459 structures. 1460 1461 workingset_refault_anon 1462 Number of refaults of previously evicted anonymous pages. 1463 1464 workingset_refault_file 1465 Number of refaults of previously evicted file pages. 1466 1467 workingset_activate_anon 1468 Number of refaulted anonymous pages that were immediately 1469 activated. 1470 1471 workingset_activate_file 1472 Number of refaulted file pages that were immediately activated. 1473 1474 workingset_restore_anon 1475 Number of restored anonymous pages which have been detected as 1476 an active workingset before they got reclaimed. 1477 1478 workingset_restore_file 1479 Number of restored file pages which have been detected as an 1480 active workingset before they got reclaimed. 1481 1482 workingset_nodereclaim 1483 Number of times a shadow node has been reclaimed 1484 1485 pgscan (npn) 1486 Amount of scanned pages (in an inactive LRU list) 1487 1488 pgsteal (npn) 1489 Amount of reclaimed pages 1490 1491 pgscan_kswapd (npn) 1492 Amount of scanned pages by kswapd (in an inactive LRU list) 1493 1494 pgscan_direct (npn) 1495 Amount of scanned pages directly (in an inactive LRU list) 1496 1497 pgscan_khugepaged (npn) 1498 Amount of scanned pages by khugepaged (in an inactive LRU list) 1499 1500 pgsteal_kswapd (npn) 1501 Amount of reclaimed pages by kswapd 1502 1503 pgsteal_direct (npn) 1504 Amount of reclaimed pages directly 1505 1506 pgsteal_khugepaged (npn) 1507 Amount of reclaimed pages by khugepaged 1508 1509 pgfault (npn) 1510 Total number of page faults incurred 1511 1512 pgmajfault (npn) 1513 Number of major page faults incurred 1514 1515 pgrefill (npn) 1516 Amount of scanned pages (in an active LRU list) 1517 1518 pgactivate (npn) 1519 Amount of pages moved to the active LRU list 1520 1521 pgdeactivate (npn) 1522 Amount of pages moved to the inactive LRU list 1523 1524 pglazyfree (npn) 1525 Amount of pages postponed to be freed under memory pressure 1526 1527 pglazyfreed (npn) 1528 Amount of reclaimed lazyfree pages 1529 1530 thp_fault_alloc (npn) 1531 Number of transparent hugepages which were allocated to satisfy 1532 a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE 1533 is not set. 1534 1535 thp_collapse_alloc (npn) 1536 Number of transparent hugepages which were allocated to allow 1537 collapsing an existing range of pages. This counter is not 1538 present when CONFIG_TRANSPARENT_HUGEPAGE is not set. 1539 1540 memory.numa_stat 1541 A read-only nested-keyed file which exists on non-root cgroups. 1542 1543 This breaks down the cgroup's memory footprint into different 1544 types of memory, type-specific details, and other information 1545 per node on the state of the memory management system. 1546 1547 This is useful for providing visibility into the NUMA locality 1548 information within an memcg since the pages are allowed to be 1549 allocated from any physical node. One of the use case is evaluating 1550 application performance by combining this information with the 1551 application's CPU allocation. 1552 1553 All memory amounts are in bytes. 1554 1555 The output format of memory.numa_stat is:: 1556 1557 type N0=<bytes in node 0> N1=<bytes in node 1> ... 1558 1559 The entries are ordered to be human readable, and new entries 1560 can show up in the middle. Don't rely on items remaining in a 1561 fixed position; use the keys to look up specific values! 1562 1563 The entries can refer to the memory.stat. 1564 1565 memory.swap.current 1566 A read-only single value file which exists on non-root 1567 cgroups. 1568 1569 The total amount of swap currently being used by the cgroup 1570 and its descendants. 1571 1572 memory.swap.high 1573 A read-write single value file which exists on non-root 1574 cgroups. The default is "max". 1575 1576 Swap usage throttle limit. If a cgroup's swap usage exceeds 1577 this limit, all its further allocations will be throttled to 1578 allow userspace to implement custom out-of-memory procedures. 1579 1580 This limit marks a point of no return for the cgroup. It is NOT 1581 designed to manage the amount of swapping a workload does 1582 during regular operation. Compare to memory.swap.max, which 1583 prohibits swapping past a set amount, but lets the cgroup 1584 continue unimpeded as long as other memory can be reclaimed. 1585 1586 Healthy workloads are not expected to reach this limit. 1587 1588 memory.swap.max 1589 A read-write single value file which exists on non-root 1590 cgroups. The default is "max". 1591 1592 Swap usage hard limit. If a cgroup's swap usage reaches this 1593 limit, anonymous memory of the cgroup will not be swapped out. 1594 1595 memory.swap.events 1596 A read-only flat-keyed file which exists on non-root cgroups. 1597 The following entries are defined. Unless specified 1598 otherwise, a value change in this file generates a file 1599 modified event. 1600 1601 high 1602 The number of times the cgroup's swap usage was over 1603 the high threshold. 1604 1605 max 1606 The number of times the cgroup's swap usage was about 1607 to go over the max boundary and swap allocation 1608 failed. 1609 1610 fail 1611 The number of times swap allocation failed either 1612 because of running out of swap system-wide or max 1613 limit. 1614 1615 When reduced under the current usage, the existing swap 1616 entries are reclaimed gradually and the swap usage may stay 1617 higher than the limit for an extended period of time. This 1618 reduces the impact on the workload and memory management. 1619 1620 memory.zswap.current 1621 A read-only single value file which exists on non-root 1622 cgroups. 1623 1624 The total amount of memory consumed by the zswap compression 1625 backend. 1626 1627 memory.zswap.max 1628 A read-write single value file which exists on non-root 1629 cgroups. The default is "max". 1630 1631 Zswap usage hard limit. If a cgroup's zswap pool reaches this 1632 limit, it will refuse to take any more stores before existing 1633 entries fault back in or are written out to disk. 1634 1635 memory.pressure 1636 A read-only nested-keyed file. 1637 1638 Shows pressure stall information for memory. See 1639 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1640 1641 1642Usage Guidelines 1643~~~~~~~~~~~~~~~~ 1644 1645"memory.high" is the main mechanism to control memory usage. 1646Over-committing on high limit (sum of high limits > available memory) 1647and letting global memory pressure to distribute memory according to 1648usage is a viable strategy. 1649 1650Because breach of the high limit doesn't trigger the OOM killer but 1651throttles the offending cgroup, a management agent has ample 1652opportunities to monitor and take appropriate actions such as granting 1653more memory or terminating the workload. 1654 1655Determining whether a cgroup has enough memory is not trivial as 1656memory usage doesn't indicate whether the workload can benefit from 1657more memory. For example, a workload which writes data received from 1658network to a file can use all available memory but can also operate as 1659performant with a small amount of memory. A measure of memory 1660pressure - how much the workload is being impacted due to lack of 1661memory - is necessary to determine whether a workload needs more 1662memory; unfortunately, memory pressure monitoring mechanism isn't 1663implemented yet. 1664 1665 1666Memory Ownership 1667~~~~~~~~~~~~~~~~ 1668 1669A memory area is charged to the cgroup which instantiated it and stays 1670charged to the cgroup until the area is released. Migrating a process 1671to a different cgroup doesn't move the memory usages that it 1672instantiated while in the previous cgroup to the new cgroup. 1673 1674A memory area may be used by processes belonging to different cgroups. 1675To which cgroup the area will be charged is in-deterministic; however, 1676over time, the memory area is likely to end up in a cgroup which has 1677enough memory allowance to avoid high reclaim pressure. 1678 1679If a cgroup sweeps a considerable amount of memory which is expected 1680to be accessed repeatedly by other cgroups, it may make sense to use 1681POSIX_FADV_DONTNEED to relinquish the ownership of memory areas 1682belonging to the affected files to ensure correct memory ownership. 1683 1684 1685IO 1686-- 1687 1688The "io" controller regulates the distribution of IO resources. This 1689controller implements both weight based and absolute bandwidth or IOPS 1690limit distribution; however, weight based distribution is available 1691only if cfq-iosched is in use and neither scheme is available for 1692blk-mq devices. 1693 1694 1695IO Interface Files 1696~~~~~~~~~~~~~~~~~~ 1697 1698 io.stat 1699 A read-only nested-keyed file. 1700 1701 Lines are keyed by $MAJ:$MIN device numbers and not ordered. 1702 The following nested keys are defined. 1703 1704 ====== ===================== 1705 rbytes Bytes read 1706 wbytes Bytes written 1707 rios Number of read IOs 1708 wios Number of write IOs 1709 dbytes Bytes discarded 1710 dios Number of discard IOs 1711 ====== ===================== 1712 1713 An example read output follows:: 1714 1715 8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0 1716 8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021 1717 1718 io.cost.qos 1719 A read-write nested-keyed file which exists only on the root 1720 cgroup. 1721 1722 This file configures the Quality of Service of the IO cost 1723 model based controller (CONFIG_BLK_CGROUP_IOCOST) which 1724 currently implements "io.weight" proportional control. Lines 1725 are keyed by $MAJ:$MIN device numbers and not ordered. The 1726 line for a given device is populated on the first write for 1727 the device on "io.cost.qos" or "io.cost.model". The following 1728 nested keys are defined. 1729 1730 ====== ===================================== 1731 enable Weight-based control enable 1732 ctrl "auto" or "user" 1733 rpct Read latency percentile [0, 100] 1734 rlat Read latency threshold 1735 wpct Write latency percentile [0, 100] 1736 wlat Write latency threshold 1737 min Minimum scaling percentage [1, 10000] 1738 max Maximum scaling percentage [1, 10000] 1739 ====== ===================================== 1740 1741 The controller is disabled by default and can be enabled by 1742 setting "enable" to 1. "rpct" and "wpct" parameters default 1743 to zero and the controller uses internal device saturation 1744 state to adjust the overall IO rate between "min" and "max". 1745 1746 When a better control quality is needed, latency QoS 1747 parameters can be configured. For example:: 1748 1749 8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0 1750 1751 shows that on sdb, the controller is enabled, will consider 1752 the device saturated if the 95th percentile of read completion 1753 latencies is above 75ms or write 150ms, and adjust the overall 1754 IO issue rate between 50% and 150% accordingly. 1755 1756 The lower the saturation point, the better the latency QoS at 1757 the cost of aggregate bandwidth. The narrower the allowed 1758 adjustment range between "min" and "max", the more conformant 1759 to the cost model the IO behavior. Note that the IO issue 1760 base rate may be far off from 100% and setting "min" and "max" 1761 blindly can lead to a significant loss of device capacity or 1762 control quality. "min" and "max" are useful for regulating 1763 devices which show wide temporary behavior changes - e.g. a 1764 ssd which accepts writes at the line speed for a while and 1765 then completely stalls for multiple seconds. 1766 1767 When "ctrl" is "auto", the parameters are controlled by the 1768 kernel and may change automatically. Setting "ctrl" to "user" 1769 or setting any of the percentile and latency parameters puts 1770 it into "user" mode and disables the automatic changes. The 1771 automatic mode can be restored by setting "ctrl" to "auto". 1772 1773 io.cost.model 1774 A read-write nested-keyed file which exists only on the root 1775 cgroup. 1776 1777 This file configures the cost model of the IO cost model based 1778 controller (CONFIG_BLK_CGROUP_IOCOST) which currently 1779 implements "io.weight" proportional control. Lines are keyed 1780 by $MAJ:$MIN device numbers and not ordered. The line for a 1781 given device is populated on the first write for the device on 1782 "io.cost.qos" or "io.cost.model". The following nested keys 1783 are defined. 1784 1785 ===== ================================ 1786 ctrl "auto" or "user" 1787 model The cost model in use - "linear" 1788 ===== ================================ 1789 1790 When "ctrl" is "auto", the kernel may change all parameters 1791 dynamically. When "ctrl" is set to "user" or any other 1792 parameters are written to, "ctrl" become "user" and the 1793 automatic changes are disabled. 1794 1795 When "model" is "linear", the following model parameters are 1796 defined. 1797 1798 ============= ======================================== 1799 [r|w]bps The maximum sequential IO throughput 1800 [r|w]seqiops The maximum 4k sequential IOs per second 1801 [r|w]randiops The maximum 4k random IOs per second 1802 ============= ======================================== 1803 1804 From the above, the builtin linear model determines the base 1805 costs of a sequential and random IO and the cost coefficient 1806 for the IO size. While simple, this model can cover most 1807 common device classes acceptably. 1808 1809 The IO cost model isn't expected to be accurate in absolute 1810 sense and is scaled to the device behavior dynamically. 1811 1812 If needed, tools/cgroup/iocost_coef_gen.py can be used to 1813 generate device-specific coefficients. 1814 1815 io.weight 1816 A read-write flat-keyed file which exists on non-root cgroups. 1817 The default is "default 100". 1818 1819 The first line is the default weight applied to devices 1820 without specific override. The rest are overrides keyed by 1821 $MAJ:$MIN device numbers and not ordered. The weights are in 1822 the range [1, 10000] and specifies the relative amount IO time 1823 the cgroup can use in relation to its siblings. 1824 1825 The default weight can be updated by writing either "default 1826 $WEIGHT" or simply "$WEIGHT". Overrides can be set by writing 1827 "$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default". 1828 1829 An example read output follows:: 1830 1831 default 100 1832 8:16 200 1833 8:0 50 1834 1835 io.max 1836 A read-write nested-keyed file which exists on non-root 1837 cgroups. 1838 1839 BPS and IOPS based IO limit. Lines are keyed by $MAJ:$MIN 1840 device numbers and not ordered. The following nested keys are 1841 defined. 1842 1843 ===== ================================== 1844 rbps Max read bytes per second 1845 wbps Max write bytes per second 1846 riops Max read IO operations per second 1847 wiops Max write IO operations per second 1848 ===== ================================== 1849 1850 When writing, any number of nested key-value pairs can be 1851 specified in any order. "max" can be specified as the value 1852 to remove a specific limit. If the same key is specified 1853 multiple times, the outcome is undefined. 1854 1855 BPS and IOPS are measured in each IO direction and IOs are 1856 delayed if limit is reached. Temporary bursts are allowed. 1857 1858 Setting read limit at 2M BPS and write at 120 IOPS for 8:16:: 1859 1860 echo "8:16 rbps=2097152 wiops=120" > io.max 1861 1862 Reading returns the following:: 1863 1864 8:16 rbps=2097152 wbps=max riops=max wiops=120 1865 1866 Write IOPS limit can be removed by writing the following:: 1867 1868 echo "8:16 wiops=max" > io.max 1869 1870 Reading now returns the following:: 1871 1872 8:16 rbps=2097152 wbps=max riops=max wiops=max 1873 1874 io.pressure 1875 A read-only nested-keyed file. 1876 1877 Shows pressure stall information for IO. See 1878 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1879 1880 1881Writeback 1882~~~~~~~~~ 1883 1884Page cache is dirtied through buffered writes and shared mmaps and 1885written asynchronously to the backing filesystem by the writeback 1886mechanism. Writeback sits between the memory and IO domains and 1887regulates the proportion of dirty memory by balancing dirtying and 1888write IOs. 1889 1890The io controller, in conjunction with the memory controller, 1891implements control of page cache writeback IOs. The memory controller 1892defines the memory domain that dirty memory ratio is calculated and 1893maintained for and the io controller defines the io domain which 1894writes out dirty pages for the memory domain. Both system-wide and 1895per-cgroup dirty memory states are examined and the more restrictive 1896of the two is enforced. 1897 1898cgroup writeback requires explicit support from the underlying 1899filesystem. Currently, cgroup writeback is implemented on ext2, ext4, 1900btrfs, f2fs, and xfs. On other filesystems, all writeback IOs are 1901attributed to the root cgroup. 1902 1903There are inherent differences in memory and writeback management 1904which affects how cgroup ownership is tracked. Memory is tracked per 1905page while writeback per inode. For the purpose of writeback, an 1906inode is assigned to a cgroup and all IO requests to write dirty pages 1907from the inode are attributed to that cgroup. 1908 1909As cgroup ownership for memory is tracked per page, there can be pages 1910which are associated with different cgroups than the one the inode is 1911associated with. These are called foreign pages. The writeback 1912constantly keeps track of foreign pages and, if a particular foreign 1913cgroup becomes the majority over a certain period of time, switches 1914the ownership of the inode to that cgroup. 1915 1916While this model is enough for most use cases where a given inode is 1917mostly dirtied by a single cgroup even when the main writing cgroup 1918changes over time, use cases where multiple cgroups write to a single 1919inode simultaneously are not supported well. In such circumstances, a 1920significant portion of IOs are likely to be attributed incorrectly. 1921As memory controller assigns page ownership on the first use and 1922doesn't update it until the page is released, even if writeback 1923strictly follows page ownership, multiple cgroups dirtying overlapping 1924areas wouldn't work as expected. It's recommended to avoid such usage 1925patterns. 1926 1927The sysctl knobs which affect writeback behavior are applied to cgroup 1928writeback as follows. 1929 1930 vm.dirty_background_ratio, vm.dirty_ratio 1931 These ratios apply the same to cgroup writeback with the 1932 amount of available memory capped by limits imposed by the 1933 memory controller and system-wide clean memory. 1934 1935 vm.dirty_background_bytes, vm.dirty_bytes 1936 For cgroup writeback, this is calculated into ratio against 1937 total available memory and applied the same way as 1938 vm.dirty[_background]_ratio. 1939 1940 1941IO Latency 1942~~~~~~~~~~ 1943 1944This is a cgroup v2 controller for IO workload protection. You provide a group 1945with a latency target, and if the average latency exceeds that target the 1946controller will throttle any peers that have a lower latency target than the 1947protected workload. 1948 1949The limits are only applied at the peer level in the hierarchy. This means that 1950in the diagram below, only groups A, B, and C will influence each other, and 1951groups D and F will influence each other. Group G will influence nobody:: 1952 1953 [root] 1954 / | \ 1955 A B C 1956 / \ | 1957 D F G 1958 1959 1960So the ideal way to configure this is to set io.latency in groups A, B, and C. 1961Generally you do not want to set a value lower than the latency your device 1962supports. Experiment to find the value that works best for your workload. 1963Start at higher than the expected latency for your device and watch the 1964avg_lat value in io.stat for your workload group to get an idea of the 1965latency you see during normal operation. Use the avg_lat value as a basis for 1966your real setting, setting at 10-15% higher than the value in io.stat. 1967 1968How IO Latency Throttling Works 1969~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1970 1971io.latency is work conserving; so as long as everybody is meeting their latency 1972target the controller doesn't do anything. Once a group starts missing its 1973target it begins throttling any peer group that has a higher target than itself. 1974This throttling takes 2 forms: 1975 1976- Queue depth throttling. This is the number of outstanding IO's a group is 1977 allowed to have. We will clamp down relatively quickly, starting at no limit 1978 and going all the way down to 1 IO at a time. 1979 1980- Artificial delay induction. There are certain types of IO that cannot be 1981 throttled without possibly adversely affecting higher priority groups. This 1982 includes swapping and metadata IO. These types of IO are allowed to occur 1983 normally, however they are "charged" to the originating group. If the 1984 originating group is being throttled you will see the use_delay and delay 1985 fields in io.stat increase. The delay value is how many microseconds that are 1986 being added to any process that runs in this group. Because this number can 1987 grow quite large if there is a lot of swapping or metadata IO occurring we 1988 limit the individual delay events to 1 second at a time. 1989 1990Once the victimized group starts meeting its latency target again it will start 1991unthrottling any peer groups that were throttled previously. If the victimized 1992group simply stops doing IO the global counter will unthrottle appropriately. 1993 1994IO Latency Interface Files 1995~~~~~~~~~~~~~~~~~~~~~~~~~~ 1996 1997 io.latency 1998 This takes a similar format as the other controllers. 1999 2000 "MAJOR:MINOR target=<target time in microseconds>" 2001 2002 io.stat 2003 If the controller is enabled you will see extra stats in io.stat in 2004 addition to the normal ones. 2005 2006 depth 2007 This is the current queue depth for the group. 2008 2009 avg_lat 2010 This is an exponential moving average with a decay rate of 1/exp 2011 bound by the sampling interval. The decay rate interval can be 2012 calculated by multiplying the win value in io.stat by the 2013 corresponding number of samples based on the win value. 2014 2015 win 2016 The sampling window size in milliseconds. This is the minimum 2017 duration of time between evaluation events. Windows only elapse 2018 with IO activity. Idle periods extend the most recent window. 2019 2020IO Priority 2021~~~~~~~~~~~ 2022 2023A single attribute controls the behavior of the I/O priority cgroup policy, 2024namely the blkio.prio.class attribute. The following values are accepted for 2025that attribute: 2026 2027 no-change 2028 Do not modify the I/O priority class. 2029 2030 none-to-rt 2031 For requests that do not have an I/O priority class (NONE), 2032 change the I/O priority class into RT. Do not modify 2033 the I/O priority class of other requests. 2034 2035 restrict-to-be 2036 For requests that do not have an I/O priority class or that have I/O 2037 priority class RT, change it into BE. Do not modify the I/O priority 2038 class of requests that have priority class IDLE. 2039 2040 idle 2041 Change the I/O priority class of all requests into IDLE, the lowest 2042 I/O priority class. 2043 2044The following numerical values are associated with the I/O priority policies: 2045 2046+-------------+---+ 2047| no-change | 0 | 2048+-------------+---+ 2049| none-to-rt | 1 | 2050+-------------+---+ 2051| rt-to-be | 2 | 2052+-------------+---+ 2053| all-to-idle | 3 | 2054+-------------+---+ 2055 2056The numerical value that corresponds to each I/O priority class is as follows: 2057 2058+-------------------------------+---+ 2059| IOPRIO_CLASS_NONE | 0 | 2060+-------------------------------+---+ 2061| IOPRIO_CLASS_RT (real-time) | 1 | 2062+-------------------------------+---+ 2063| IOPRIO_CLASS_BE (best effort) | 2 | 2064+-------------------------------+---+ 2065| IOPRIO_CLASS_IDLE | 3 | 2066+-------------------------------+---+ 2067 2068The algorithm to set the I/O priority class for a request is as follows: 2069 2070- Translate the I/O priority class policy into a number. 2071- Change the request I/O priority class into the maximum of the I/O priority 2072 class policy number and the numerical I/O priority class. 2073 2074PID 2075--- 2076 2077The process number controller is used to allow a cgroup to stop any 2078new tasks from being fork()'d or clone()'d after a specified limit is 2079reached. 2080 2081The number of tasks in a cgroup can be exhausted in ways which other 2082controllers cannot prevent, thus warranting its own controller. For 2083example, a fork bomb is likely to exhaust the number of tasks before 2084hitting memory restrictions. 2085 2086Note that PIDs used in this controller refer to TIDs, process IDs as 2087used by the kernel. 2088 2089 2090PID Interface Files 2091~~~~~~~~~~~~~~~~~~~ 2092 2093 pids.max 2094 A read-write single value file which exists on non-root 2095 cgroups. The default is "max". 2096 2097 Hard limit of number of processes. 2098 2099 pids.current 2100 A read-only single value file which exists on all cgroups. 2101 2102 The number of processes currently in the cgroup and its 2103 descendants. 2104 2105Organisational operations are not blocked by cgroup policies, so it is 2106possible to have pids.current > pids.max. This can be done by either 2107setting the limit to be smaller than pids.current, or attaching enough 2108processes to the cgroup such that pids.current is larger than 2109pids.max. However, it is not possible to violate a cgroup PID policy 2110through fork() or clone(). These will return -EAGAIN if the creation 2111of a new process would cause a cgroup policy to be violated. 2112 2113 2114Cpuset 2115------ 2116 2117The "cpuset" controller provides a mechanism for constraining 2118the CPU and memory node placement of tasks to only the resources 2119specified in the cpuset interface files in a task's current cgroup. 2120This is especially valuable on large NUMA systems where placing jobs 2121on properly sized subsets of the systems with careful processor and 2122memory placement to reduce cross-node memory access and contention 2123can improve overall system performance. 2124 2125The "cpuset" controller is hierarchical. That means the controller 2126cannot use CPUs or memory nodes not allowed in its parent. 2127 2128 2129Cpuset Interface Files 2130~~~~~~~~~~~~~~~~~~~~~~ 2131 2132 cpuset.cpus 2133 A read-write multiple values file which exists on non-root 2134 cpuset-enabled cgroups. 2135 2136 It lists the requested CPUs to be used by tasks within this 2137 cgroup. The actual list of CPUs to be granted, however, is 2138 subjected to constraints imposed by its parent and can differ 2139 from the requested CPUs. 2140 2141 The CPU numbers are comma-separated numbers or ranges. 2142 For example:: 2143 2144 # cat cpuset.cpus 2145 0-4,6,8-10 2146 2147 An empty value indicates that the cgroup is using the same 2148 setting as the nearest cgroup ancestor with a non-empty 2149 "cpuset.cpus" or all the available CPUs if none is found. 2150 2151 The value of "cpuset.cpus" stays constant until the next update 2152 and won't be affected by any CPU hotplug events. 2153 2154 cpuset.cpus.effective 2155 A read-only multiple values file which exists on all 2156 cpuset-enabled cgroups. 2157 2158 It lists the onlined CPUs that are actually granted to this 2159 cgroup by its parent. These CPUs are allowed to be used by 2160 tasks within the current cgroup. 2161 2162 If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows 2163 all the CPUs from the parent cgroup that can be available to 2164 be used by this cgroup. Otherwise, it should be a subset of 2165 "cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus" 2166 can be granted. In this case, it will be treated just like an 2167 empty "cpuset.cpus". 2168 2169 Its value will be affected by CPU hotplug events. 2170 2171 cpuset.mems 2172 A read-write multiple values file which exists on non-root 2173 cpuset-enabled cgroups. 2174 2175 It lists the requested memory nodes to be used by tasks within 2176 this cgroup. The actual list of memory nodes granted, however, 2177 is subjected to constraints imposed by its parent and can differ 2178 from the requested memory nodes. 2179 2180 The memory node numbers are comma-separated numbers or ranges. 2181 For example:: 2182 2183 # cat cpuset.mems 2184 0-1,3 2185 2186 An empty value indicates that the cgroup is using the same 2187 setting as the nearest cgroup ancestor with a non-empty 2188 "cpuset.mems" or all the available memory nodes if none 2189 is found. 2190 2191 The value of "cpuset.mems" stays constant until the next update 2192 and won't be affected by any memory nodes hotplug events. 2193 2194 Setting a non-empty value to "cpuset.mems" causes memory of 2195 tasks within the cgroup to be migrated to the designated nodes if 2196 they are currently using memory outside of the designated nodes. 2197 2198 There is a cost for this memory migration. The migration 2199 may not be complete and some memory pages may be left behind. 2200 So it is recommended that "cpuset.mems" should be set properly 2201 before spawning new tasks into the cpuset. Even if there is 2202 a need to change "cpuset.mems" with active tasks, it shouldn't 2203 be done frequently. 2204 2205 cpuset.mems.effective 2206 A read-only multiple values file which exists on all 2207 cpuset-enabled cgroups. 2208 2209 It lists the onlined memory nodes that are actually granted to 2210 this cgroup by its parent. These memory nodes are allowed to 2211 be used by tasks within the current cgroup. 2212 2213 If "cpuset.mems" is empty, it shows all the memory nodes from the 2214 parent cgroup that will be available to be used by this cgroup. 2215 Otherwise, it should be a subset of "cpuset.mems" unless none of 2216 the memory nodes listed in "cpuset.mems" can be granted. In this 2217 case, it will be treated just like an empty "cpuset.mems". 2218 2219 Its value will be affected by memory nodes hotplug events. 2220 2221 cpuset.cpus.partition 2222 A read-write single value file which exists on non-root 2223 cpuset-enabled cgroups. This flag is owned by the parent cgroup 2224 and is not delegatable. 2225 2226 It accepts only the following input values when written to. 2227 2228 ========== ===================================== 2229 "member" Non-root member of a partition 2230 "root" Partition root 2231 "isolated" Partition root without load balancing 2232 ========== ===================================== 2233 2234 The root cgroup is always a partition root and its state 2235 cannot be changed. All other non-root cgroups start out as 2236 "member". 2237 2238 When set to "root", the current cgroup is the root of a new 2239 partition or scheduling domain that comprises itself and all 2240 its descendants except those that are separate partition roots 2241 themselves and their descendants. 2242 2243 When set to "isolated", the CPUs in that partition root will 2244 be in an isolated state without any load balancing from the 2245 scheduler. Tasks placed in such a partition with multiple 2246 CPUs should be carefully distributed and bound to each of the 2247 individual CPUs for optimal performance. 2248 2249 The value shown in "cpuset.cpus.effective" of a partition root 2250 is the CPUs that the partition root can dedicate to a potential 2251 new child partition root. The new child subtracts available 2252 CPUs from its parent "cpuset.cpus.effective". 2253 2254 A partition root ("root" or "isolated") can be in one of the 2255 two possible states - valid or invalid. An invalid partition 2256 root is in a degraded state where some state information may 2257 be retained, but behaves more like a "member". 2258 2259 All possible state transitions among "member", "root" and 2260 "isolated" are allowed. 2261 2262 On read, the "cpuset.cpus.partition" file can show the following 2263 values. 2264 2265 ============================= ===================================== 2266 "member" Non-root member of a partition 2267 "root" Partition root 2268 "isolated" Partition root without load balancing 2269 "root invalid (<reason>)" Invalid partition root 2270 "isolated invalid (<reason>)" Invalid isolated partition root 2271 ============================= ===================================== 2272 2273 In the case of an invalid partition root, a descriptive string on 2274 why the partition is invalid is included within parentheses. 2275 2276 For a partition root to become valid, the following conditions 2277 must be met. 2278 2279 1) The "cpuset.cpus" is exclusive with its siblings , i.e. they 2280 are not shared by any of its siblings (exclusivity rule). 2281 2) The parent cgroup is a valid partition root. 2282 3) The "cpuset.cpus" is not empty and must contain at least 2283 one of the CPUs from parent's "cpuset.cpus", i.e. they overlap. 2284 4) The "cpuset.cpus.effective" cannot be empty unless there is 2285 no task associated with this partition. 2286 2287 External events like hotplug or changes to "cpuset.cpus" can 2288 cause a valid partition root to become invalid and vice versa. 2289 Note that a task cannot be moved to a cgroup with empty 2290 "cpuset.cpus.effective". 2291 2292 For a valid partition root with the sibling cpu exclusivity 2293 rule enabled, changes made to "cpuset.cpus" that violate the 2294 exclusivity rule will invalidate the partition as well as its 2295 sibiling partitions with conflicting cpuset.cpus values. So 2296 care must be taking in changing "cpuset.cpus". 2297 2298 A valid non-root parent partition may distribute out all its CPUs 2299 to its child partitions when there is no task associated with it. 2300 2301 Care must be taken to change a valid partition root to 2302 "member" as all its child partitions, if present, will become 2303 invalid causing disruption to tasks running in those child 2304 partitions. These inactivated partitions could be recovered if 2305 their parent is switched back to a partition root with a proper 2306 set of "cpuset.cpus". 2307 2308 Poll and inotify events are triggered whenever the state of 2309 "cpuset.cpus.partition" changes. That includes changes caused 2310 by write to "cpuset.cpus.partition", cpu hotplug or other 2311 changes that modify the validity status of the partition. 2312 This will allow user space agents to monitor unexpected changes 2313 to "cpuset.cpus.partition" without the need to do continuous 2314 polling. 2315 2316 2317Device controller 2318----------------- 2319 2320Device controller manages access to device files. It includes both 2321creation of new device files (using mknod), and access to the 2322existing device files. 2323 2324Cgroup v2 device controller has no interface files and is implemented 2325on top of cgroup BPF. To control access to device files, a user may 2326create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach 2327them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a 2328device file, corresponding BPF programs will be executed, and depending 2329on the return value the attempt will succeed or fail with -EPERM. 2330 2331A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the 2332bpf_cgroup_dev_ctx structure, which describes the device access attempt: 2333access type (mknod/read/write) and device (type, major and minor numbers). 2334If the program returns 0, the attempt fails with -EPERM, otherwise it 2335succeeds. 2336 2337An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in 2338tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree. 2339 2340 2341RDMA 2342---- 2343 2344The "rdma" controller regulates the distribution and accounting of 2345RDMA resources. 2346 2347RDMA Interface Files 2348~~~~~~~~~~~~~~~~~~~~ 2349 2350 rdma.max 2351 A readwrite nested-keyed file that exists for all the cgroups 2352 except root that describes current configured resource limit 2353 for a RDMA/IB device. 2354 2355 Lines are keyed by device name and are not ordered. 2356 Each line contains space separated resource name and its configured 2357 limit that can be distributed. 2358 2359 The following nested keys are defined. 2360 2361 ========== ============================= 2362 hca_handle Maximum number of HCA Handles 2363 hca_object Maximum number of HCA Objects 2364 ========== ============================= 2365 2366 An example for mlx4 and ocrdma device follows:: 2367 2368 mlx4_0 hca_handle=2 hca_object=2000 2369 ocrdma1 hca_handle=3 hca_object=max 2370 2371 rdma.current 2372 A read-only file that describes current resource usage. 2373 It exists for all the cgroup except root. 2374 2375 An example for mlx4 and ocrdma device follows:: 2376 2377 mlx4_0 hca_handle=1 hca_object=20 2378 ocrdma1 hca_handle=1 hca_object=23 2379 2380HugeTLB 2381------- 2382 2383The HugeTLB controller allows to limit the HugeTLB usage per control group and 2384enforces the controller limit during page fault. 2385 2386HugeTLB Interface Files 2387~~~~~~~~~~~~~~~~~~~~~~~ 2388 2389 hugetlb.<hugepagesize>.current 2390 Show current usage for "hugepagesize" hugetlb. It exists for all 2391 the cgroup except root. 2392 2393 hugetlb.<hugepagesize>.max 2394 Set/show the hard limit of "hugepagesize" hugetlb usage. 2395 The default value is "max". It exists for all the cgroup except root. 2396 2397 hugetlb.<hugepagesize>.events 2398 A read-only flat-keyed file which exists on non-root cgroups. 2399 2400 max 2401 The number of allocation failure due to HugeTLB limit 2402 2403 hugetlb.<hugepagesize>.events.local 2404 Similar to hugetlb.<hugepagesize>.events but the fields in the file 2405 are local to the cgroup i.e. not hierarchical. The file modified event 2406 generated on this file reflects only the local events. 2407 2408 hugetlb.<hugepagesize>.numa_stat 2409 Similar to memory.numa_stat, it shows the numa information of the 2410 hugetlb pages of <hugepagesize> in this cgroup. Only active in 2411 use hugetlb pages are included. The per-node values are in bytes. 2412 2413Misc 2414---- 2415 2416The Miscellaneous cgroup provides the resource limiting and tracking 2417mechanism for the scalar resources which cannot be abstracted like the other 2418cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config 2419option. 2420 2421A resource can be added to the controller via enum misc_res_type{} in the 2422include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[] 2423in the kernel/cgroup/misc.c file. Provider of the resource must set its 2424capacity prior to using the resource by calling misc_cg_set_capacity(). 2425 2426Once a capacity is set then the resource usage can be updated using charge and 2427uncharge APIs. All of the APIs to interact with misc controller are in 2428include/linux/misc_cgroup.h. 2429 2430Misc Interface Files 2431~~~~~~~~~~~~~~~~~~~~ 2432 2433Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then: 2434 2435 misc.capacity 2436 A read-only flat-keyed file shown only in the root cgroup. It shows 2437 miscellaneous scalar resources available on the platform along with 2438 their quantities:: 2439 2440 $ cat misc.capacity 2441 res_a 50 2442 res_b 10 2443 2444 misc.current 2445 A read-only flat-keyed file shown in the non-root cgroups. It shows 2446 the current usage of the resources in the cgroup and its children.:: 2447 2448 $ cat misc.current 2449 res_a 3 2450 res_b 0 2451 2452 misc.max 2453 A read-write flat-keyed file shown in the non root cgroups. Allowed 2454 maximum usage of the resources in the cgroup and its children.:: 2455 2456 $ cat misc.max 2457 res_a max 2458 res_b 4 2459 2460 Limit can be set by:: 2461 2462 # echo res_a 1 > misc.max 2463 2464 Limit can be set to max by:: 2465 2466 # echo res_a max > misc.max 2467 2468 Limits can be set higher than the capacity value in the misc.capacity 2469 file. 2470 2471 misc.events 2472 A read-only flat-keyed file which exists on non-root cgroups. The 2473 following entries are defined. Unless specified otherwise, a value 2474 change in this file generates a file modified event. All fields in 2475 this file are hierarchical. 2476 2477 max 2478 The number of times the cgroup's resource usage was 2479 about to go over the max boundary. 2480 2481Migration and Ownership 2482~~~~~~~~~~~~~~~~~~~~~~~ 2483 2484A miscellaneous scalar resource is charged to the cgroup in which it is used 2485first, and stays charged to that cgroup until that resource is freed. Migrating 2486a process to a different cgroup does not move the charge to the destination 2487cgroup where the process has moved. 2488 2489Others 2490------ 2491 2492perf_event 2493~~~~~~~~~~ 2494 2495perf_event controller, if not mounted on a legacy hierarchy, is 2496automatically enabled on the v2 hierarchy so that perf events can 2497always be filtered by cgroup v2 path. The controller can still be 2498moved to a legacy hierarchy after v2 hierarchy is populated. 2499 2500 2501Non-normative information 2502------------------------- 2503 2504This section contains information that isn't considered to be a part of 2505the stable kernel API and so is subject to change. 2506 2507 2508CPU controller root cgroup process behaviour 2509~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2510 2511When distributing CPU cycles in the root cgroup each thread in this 2512cgroup is treated as if it was hosted in a separate child cgroup of the 2513root cgroup. This child cgroup weight is dependent on its thread nice 2514level. 2515 2516For details of this mapping see sched_prio_to_weight array in 2517kernel/sched/core.c file (values from this array should be scaled 2518appropriately so the neutral - nice 0 - value is 100 instead of 1024). 2519 2520 2521IO controller root cgroup process behaviour 2522~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2523 2524Root cgroup processes are hosted in an implicit leaf child node. 2525When distributing IO resources this implicit child node is taken into 2526account as if it was a normal child cgroup of the root cgroup with a 2527weight value of 200. 2528 2529 2530Namespace 2531========= 2532 2533Basics 2534------ 2535 2536cgroup namespace provides a mechanism to virtualize the view of the 2537"/proc/$PID/cgroup" file and cgroup mounts. The CLONE_NEWCGROUP clone 2538flag can be used with clone(2) and unshare(2) to create a new cgroup 2539namespace. The process running inside the cgroup namespace will have 2540its "/proc/$PID/cgroup" output restricted to cgroupns root. The 2541cgroupns root is the cgroup of the process at the time of creation of 2542the cgroup namespace. 2543 2544Without cgroup namespace, the "/proc/$PID/cgroup" file shows the 2545complete path of the cgroup of a process. In a container setup where 2546a set of cgroups and namespaces are intended to isolate processes the 2547"/proc/$PID/cgroup" file may leak potential system level information 2548to the isolated processes. For example:: 2549 2550 # cat /proc/self/cgroup 2551 0::/batchjobs/container_id1 2552 2553The path '/batchjobs/container_id1' can be considered as system-data 2554and undesirable to expose to the isolated processes. cgroup namespace 2555can be used to restrict visibility of this path. For example, before 2556creating a cgroup namespace, one would see:: 2557 2558 # ls -l /proc/self/ns/cgroup 2559 lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835] 2560 # cat /proc/self/cgroup 2561 0::/batchjobs/container_id1 2562 2563After unsharing a new namespace, the view changes:: 2564 2565 # ls -l /proc/self/ns/cgroup 2566 lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183] 2567 # cat /proc/self/cgroup 2568 0::/ 2569 2570When some thread from a multi-threaded process unshares its cgroup 2571namespace, the new cgroupns gets applied to the entire process (all 2572the threads). This is natural for the v2 hierarchy; however, for the 2573legacy hierarchies, this may be unexpected. 2574 2575A cgroup namespace is alive as long as there are processes inside or 2576mounts pinning it. When the last usage goes away, the cgroup 2577namespace is destroyed. The cgroupns root and the actual cgroups 2578remain. 2579 2580 2581The Root and Views 2582------------------ 2583 2584The 'cgroupns root' for a cgroup namespace is the cgroup in which the 2585process calling unshare(2) is running. For example, if a process in 2586/batchjobs/container_id1 cgroup calls unshare, cgroup 2587/batchjobs/container_id1 becomes the cgroupns root. For the 2588init_cgroup_ns, this is the real root ('/') cgroup. 2589 2590The cgroupns root cgroup does not change even if the namespace creator 2591process later moves to a different cgroup:: 2592 2593 # ~/unshare -c # unshare cgroupns in some cgroup 2594 # cat /proc/self/cgroup 2595 0::/ 2596 # mkdir sub_cgrp_1 2597 # echo 0 > sub_cgrp_1/cgroup.procs 2598 # cat /proc/self/cgroup 2599 0::/sub_cgrp_1 2600 2601Each process gets its namespace-specific view of "/proc/$PID/cgroup" 2602 2603Processes running inside the cgroup namespace will be able to see 2604cgroup paths (in /proc/self/cgroup) only inside their root cgroup. 2605From within an unshared cgroupns:: 2606 2607 # sleep 100000 & 2608 [1] 7353 2609 # echo 7353 > sub_cgrp_1/cgroup.procs 2610 # cat /proc/7353/cgroup 2611 0::/sub_cgrp_1 2612 2613From the initial cgroup namespace, the real cgroup path will be 2614visible:: 2615 2616 $ cat /proc/7353/cgroup 2617 0::/batchjobs/container_id1/sub_cgrp_1 2618 2619From a sibling cgroup namespace (that is, a namespace rooted at a 2620different cgroup), the cgroup path relative to its own cgroup 2621namespace root will be shown. For instance, if PID 7353's cgroup 2622namespace root is at '/batchjobs/container_id2', then it will see:: 2623 2624 # cat /proc/7353/cgroup 2625 0::/../container_id2/sub_cgrp_1 2626 2627Note that the relative path always starts with '/' to indicate that 2628its relative to the cgroup namespace root of the caller. 2629 2630 2631Migration and setns(2) 2632---------------------- 2633 2634Processes inside a cgroup namespace can move into and out of the 2635namespace root if they have proper access to external cgroups. For 2636example, from inside a namespace with cgroupns root at 2637/batchjobs/container_id1, and assuming that the global hierarchy is 2638still accessible inside cgroupns:: 2639 2640 # cat /proc/7353/cgroup 2641 0::/sub_cgrp_1 2642 # echo 7353 > batchjobs/container_id2/cgroup.procs 2643 # cat /proc/7353/cgroup 2644 0::/../container_id2 2645 2646Note that this kind of setup is not encouraged. A task inside cgroup 2647namespace should only be exposed to its own cgroupns hierarchy. 2648 2649setns(2) to another cgroup namespace is allowed when: 2650 2651(a) the process has CAP_SYS_ADMIN against its current user namespace 2652(b) the process has CAP_SYS_ADMIN against the target cgroup 2653 namespace's userns 2654 2655No implicit cgroup changes happen with attaching to another cgroup 2656namespace. It is expected that the someone moves the attaching 2657process under the target cgroup namespace root. 2658 2659 2660Interaction with Other Namespaces 2661--------------------------------- 2662 2663Namespace specific cgroup hierarchy can be mounted by a process 2664running inside a non-init cgroup namespace:: 2665 2666 # mount -t cgroup2 none $MOUNT_POINT 2667 2668This will mount the unified cgroup hierarchy with cgroupns root as the 2669filesystem root. The process needs CAP_SYS_ADMIN against its user and 2670mount namespaces. 2671 2672The virtualization of /proc/self/cgroup file combined with restricting 2673the view of cgroup hierarchy by namespace-private cgroupfs mount 2674provides a properly isolated cgroup view inside the container. 2675 2676 2677Information on Kernel Programming 2678================================= 2679 2680This section contains kernel programming information in the areas 2681where interacting with cgroup is necessary. cgroup core and 2682controllers are not covered. 2683 2684 2685Filesystem Support for Writeback 2686-------------------------------- 2687 2688A filesystem can support cgroup writeback by updating 2689address_space_operations->writepage[s]() to annotate bio's using the 2690following two functions. 2691 2692 wbc_init_bio(@wbc, @bio) 2693 Should be called for each bio carrying writeback data and 2694 associates the bio with the inode's owner cgroup and the 2695 corresponding request queue. This must be called after 2696 a queue (device) has been associated with the bio and 2697 before submission. 2698 2699 wbc_account_cgroup_owner(@wbc, @page, @bytes) 2700 Should be called for each data segment being written out. 2701 While this function doesn't care exactly when it's called 2702 during the writeback session, it's the easiest and most 2703 natural to call it as data segments are added to a bio. 2704 2705With writeback bio's annotated, cgroup support can be enabled per 2706super_block by setting SB_I_CGROUPWB in ->s_iflags. This allows for 2707selective disabling of cgroup writeback support which is helpful when 2708certain filesystem features, e.g. journaled data mode, are 2709incompatible. 2710 2711wbc_init_bio() binds the specified bio to its cgroup. Depending on 2712the configuration, the bio may be executed at a lower priority and if 2713the writeback session is holding shared resources, e.g. a journal 2714entry, may lead to priority inversion. There is no one easy solution 2715for the problem. Filesystems can try to work around specific problem 2716cases by skipping wbc_init_bio() and using bio_associate_blkg() 2717directly. 2718 2719 2720Deprecated v1 Core Features 2721=========================== 2722 2723- Multiple hierarchies including named ones are not supported. 2724 2725- All v1 mount options are not supported. 2726 2727- The "tasks" file is removed and "cgroup.procs" is not sorted. 2728 2729- "cgroup.clone_children" is removed. 2730 2731- /proc/cgroups is meaningless for v2. Use "cgroup.controllers" file 2732 at the root instead. 2733 2734 2735Issues with v1 and Rationales for v2 2736==================================== 2737 2738Multiple Hierarchies 2739-------------------- 2740 2741cgroup v1 allowed an arbitrary number of hierarchies and each 2742hierarchy could host any number of controllers. While this seemed to 2743provide a high level of flexibility, it wasn't useful in practice. 2744 2745For example, as there is only one instance of each controller, utility 2746type controllers such as freezer which can be useful in all 2747hierarchies could only be used in one. The issue is exacerbated by 2748the fact that controllers couldn't be moved to another hierarchy once 2749hierarchies were populated. Another issue was that all controllers 2750bound to a hierarchy were forced to have exactly the same view of the 2751hierarchy. It wasn't possible to vary the granularity depending on 2752the specific controller. 2753 2754In practice, these issues heavily limited which controllers could be 2755put on the same hierarchy and most configurations resorted to putting 2756each controller on its own hierarchy. Only closely related ones, such 2757as the cpu and cpuacct controllers, made sense to be put on the same 2758hierarchy. This often meant that userland ended up managing multiple 2759similar hierarchies repeating the same steps on each hierarchy 2760whenever a hierarchy management operation was necessary. 2761 2762Furthermore, support for multiple hierarchies came at a steep cost. 2763It greatly complicated cgroup core implementation but more importantly 2764the support for multiple hierarchies restricted how cgroup could be 2765used in general and what controllers was able to do. 2766 2767There was no limit on how many hierarchies there might be, which meant 2768that a thread's cgroup membership couldn't be described in finite 2769length. The key might contain any number of entries and was unlimited 2770in length, which made it highly awkward to manipulate and led to 2771addition of controllers which existed only to identify membership, 2772which in turn exacerbated the original problem of proliferating number 2773of hierarchies. 2774 2775Also, as a controller couldn't have any expectation regarding the 2776topologies of hierarchies other controllers might be on, each 2777controller had to assume that all other controllers were attached to 2778completely orthogonal hierarchies. This made it impossible, or at 2779least very cumbersome, for controllers to cooperate with each other. 2780 2781In most use cases, putting controllers on hierarchies which are 2782completely orthogonal to each other isn't necessary. What usually is 2783called for is the ability to have differing levels of granularity 2784depending on the specific controller. In other words, hierarchy may 2785be collapsed from leaf towards root when viewed from specific 2786controllers. For example, a given configuration might not care about 2787how memory is distributed beyond a certain level while still wanting 2788to control how CPU cycles are distributed. 2789 2790 2791Thread Granularity 2792------------------ 2793 2794cgroup v1 allowed threads of a process to belong to different cgroups. 2795This didn't make sense for some controllers and those controllers 2796ended up implementing different ways to ignore such situations but 2797much more importantly it blurred the line between API exposed to 2798individual applications and system management interface. 2799 2800Generally, in-process knowledge is available only to the process 2801itself; thus, unlike service-level organization of processes, 2802categorizing threads of a process requires active participation from 2803the application which owns the target process. 2804 2805cgroup v1 had an ambiguously defined delegation model which got abused 2806in combination with thread granularity. cgroups were delegated to 2807individual applications so that they can create and manage their own 2808sub-hierarchies and control resource distributions along them. This 2809effectively raised cgroup to the status of a syscall-like API exposed 2810to lay programs. 2811 2812First of all, cgroup has a fundamentally inadequate interface to be 2813exposed this way. For a process to access its own knobs, it has to 2814extract the path on the target hierarchy from /proc/self/cgroup, 2815construct the path by appending the name of the knob to the path, open 2816and then read and/or write to it. This is not only extremely clunky 2817and unusual but also inherently racy. There is no conventional way to 2818define transaction across the required steps and nothing can guarantee 2819that the process would actually be operating on its own sub-hierarchy. 2820 2821cgroup controllers implemented a number of knobs which would never be 2822accepted as public APIs because they were just adding control knobs to 2823system-management pseudo filesystem. cgroup ended up with interface 2824knobs which were not properly abstracted or refined and directly 2825revealed kernel internal details. These knobs got exposed to 2826individual applications through the ill-defined delegation mechanism 2827effectively abusing cgroup as a shortcut to implementing public APIs 2828without going through the required scrutiny. 2829 2830This was painful for both userland and kernel. Userland ended up with 2831misbehaving and poorly abstracted interfaces and kernel exposing and 2832locked into constructs inadvertently. 2833 2834 2835Competition Between Inner Nodes and Threads 2836------------------------------------------- 2837 2838cgroup v1 allowed threads to be in any cgroups which created an 2839interesting problem where threads belonging to a parent cgroup and its 2840children cgroups competed for resources. This was nasty as two 2841different types of entities competed and there was no obvious way to 2842settle it. Different controllers did different things. 2843 2844The cpu controller considered threads and cgroups as equivalents and 2845mapped nice levels to cgroup weights. This worked for some cases but 2846fell flat when children wanted to be allocated specific ratios of CPU 2847cycles and the number of internal threads fluctuated - the ratios 2848constantly changed as the number of competing entities fluctuated. 2849There also were other issues. The mapping from nice level to weight 2850wasn't obvious or universal, and there were various other knobs which 2851simply weren't available for threads. 2852 2853The io controller implicitly created a hidden leaf node for each 2854cgroup to host the threads. The hidden leaf had its own copies of all 2855the knobs with ``leaf_`` prefixed. While this allowed equivalent 2856control over internal threads, it was with serious drawbacks. It 2857always added an extra layer of nesting which wouldn't be necessary 2858otherwise, made the interface messy and significantly complicated the 2859implementation. 2860 2861The memory controller didn't have a way to control what happened 2862between internal tasks and child cgroups and the behavior was not 2863clearly defined. There were attempts to add ad-hoc behaviors and 2864knobs to tailor the behavior to specific workloads which would have 2865led to problems extremely difficult to resolve in the long term. 2866 2867Multiple controllers struggled with internal tasks and came up with 2868different ways to deal with it; unfortunately, all the approaches were 2869severely flawed and, furthermore, the widely different behaviors 2870made cgroup as a whole highly inconsistent. 2871 2872This clearly is a problem which needs to be addressed from cgroup core 2873in a uniform way. 2874 2875 2876Other Interface Issues 2877---------------------- 2878 2879cgroup v1 grew without oversight and developed a large number of 2880idiosyncrasies and inconsistencies. One issue on the cgroup core side 2881was how an empty cgroup was notified - a userland helper binary was 2882forked and executed for each event. The event delivery wasn't 2883recursive or delegatable. The limitations of the mechanism also led 2884to in-kernel event delivery filtering mechanism further complicating 2885the interface. 2886 2887Controller interfaces were problematic too. An extreme example is 2888controllers completely ignoring hierarchical organization and treating 2889all cgroups as if they were all located directly under the root 2890cgroup. Some controllers exposed a large amount of inconsistent 2891implementation details to userland. 2892 2893There also was no consistency across controllers. When a new cgroup 2894was created, some controllers defaulted to not imposing extra 2895restrictions while others disallowed any resource usage until 2896explicitly configured. Configuration knobs for the same type of 2897control used widely differing naming schemes and formats. Statistics 2898and information knobs were named arbitrarily and used different 2899formats and units even in the same controller. 2900 2901cgroup v2 establishes common conventions where appropriate and updates 2902controllers so that they expose minimal and consistent interfaces. 2903 2904 2905Controller Issues and Remedies 2906------------------------------ 2907 2908Memory 2909~~~~~~ 2910 2911The original lower boundary, the soft limit, is defined as a limit 2912that is per default unset. As a result, the set of cgroups that 2913global reclaim prefers is opt-in, rather than opt-out. The costs for 2914optimizing these mostly negative lookups are so high that the 2915implementation, despite its enormous size, does not even provide the 2916basic desirable behavior. First off, the soft limit has no 2917hierarchical meaning. All configured groups are organized in a global 2918rbtree and treated like equal peers, regardless where they are located 2919in the hierarchy. This makes subtree delegation impossible. Second, 2920the soft limit reclaim pass is so aggressive that it not just 2921introduces high allocation latencies into the system, but also impacts 2922system performance due to overreclaim, to the point where the feature 2923becomes self-defeating. 2924 2925The memory.low boundary on the other hand is a top-down allocated 2926reserve. A cgroup enjoys reclaim protection when it's within its 2927effective low, which makes delegation of subtrees possible. It also 2928enjoys having reclaim pressure proportional to its overage when 2929above its effective low. 2930 2931The original high boundary, the hard limit, is defined as a strict 2932limit that can not budge, even if the OOM killer has to be called. 2933But this generally goes against the goal of making the most out of the 2934available memory. The memory consumption of workloads varies during 2935runtime, and that requires users to overcommit. But doing that with a 2936strict upper limit requires either a fairly accurate prediction of the 2937working set size or adding slack to the limit. Since working set size 2938estimation is hard and error prone, and getting it wrong results in 2939OOM kills, most users tend to err on the side of a looser limit and 2940end up wasting precious resources. 2941 2942The memory.high boundary on the other hand can be set much more 2943conservatively. When hit, it throttles allocations by forcing them 2944into direct reclaim to work off the excess, but it never invokes the 2945OOM killer. As a result, a high boundary that is chosen too 2946aggressively will not terminate the processes, but instead it will 2947lead to gradual performance degradation. The user can monitor this 2948and make corrections until the minimal memory footprint that still 2949gives acceptable performance is found. 2950 2951In extreme cases, with many concurrent allocations and a complete 2952breakdown of reclaim progress within the group, the high boundary can 2953be exceeded. But even then it's mostly better to satisfy the 2954allocation from the slack available in other groups or the rest of the 2955system than killing the group. Otherwise, memory.max is there to 2956limit this type of spillover and ultimately contain buggy or even 2957malicious applications. 2958 2959Setting the original memory.limit_in_bytes below the current usage was 2960subject to a race condition, where concurrent charges could cause the 2961limit setting to fail. memory.max on the other hand will first set the 2962limit to prevent new charges, and then reclaim and OOM kill until the 2963new limit is met - or the task writing to memory.max is killed. 2964 2965The combined memory+swap accounting and limiting is replaced by real 2966control over swap space. 2967 2968The main argument for a combined memory+swap facility in the original 2969cgroup design was that global or parental pressure would always be 2970able to swap all anonymous memory of a child group, regardless of the 2971child's own (possibly untrusted) configuration. However, untrusted 2972groups can sabotage swapping by other means - such as referencing its 2973anonymous memory in a tight loop - and an admin can not assume full 2974swappability when overcommitting untrusted jobs. 2975 2976For trusted jobs, on the other hand, a combined counter is not an 2977intuitive userspace interface, and it flies in the face of the idea 2978that cgroup controllers should account and limit specific physical 2979resources. Swap space is a resource like all others in the system, 2980and that's why unified hierarchy allows distributing it separately. 2981