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 up to 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 up to 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 up to 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 up to $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. If a cgroup's usage goes 1217 over the high boundary, the processes of the cgroup are 1218 throttled and put under heavy reclaim pressure. 1219 1220 Going over the high limit never invokes the OOM killer and 1221 under extreme conditions the limit may be breached. The high 1222 limit should be used in scenarios where an external process 1223 monitors the limited cgroup to alleviate heavy reclaim 1224 pressure. 1225 1226 memory.max 1227 A read-write single value file which exists on non-root 1228 cgroups. The default is "max". 1229 1230 Memory usage hard limit. This is the main mechanism to limit 1231 memory usage of a cgroup. If a cgroup's memory usage reaches 1232 this limit and can't be reduced, the OOM killer is invoked in 1233 the cgroup. Under certain circumstances, the usage may go 1234 over the limit temporarily. 1235 1236 In default configuration regular 0-order allocations always 1237 succeed unless OOM killer chooses current task as a victim. 1238 1239 Some kinds of allocations don't invoke the OOM killer. 1240 Caller could retry them differently, return into userspace 1241 as -ENOMEM or silently ignore in cases like disk readahead. 1242 1243 memory.reclaim 1244 A write-only nested-keyed file which exists for all cgroups. 1245 1246 This is a simple interface to trigger memory reclaim in the 1247 target cgroup. 1248 1249 This file accepts a single key, the number of bytes to reclaim. 1250 No nested keys are currently supported. 1251 1252 Example:: 1253 1254 echo "1G" > memory.reclaim 1255 1256 The interface can be later extended with nested keys to 1257 configure the reclaim behavior. For example, specify the 1258 type of memory to reclaim from (anon, file, ..). 1259 1260 Please note that the kernel can over or under reclaim from 1261 the target cgroup. If less bytes are reclaimed than the 1262 specified amount, -EAGAIN is returned. 1263 1264 Please note that the proactive reclaim (triggered by this 1265 interface) is not meant to indicate memory pressure on the 1266 memory cgroup. Therefore socket memory balancing triggered by 1267 the memory reclaim normally is not exercised in this case. 1268 This means that the networking layer will not adapt based on 1269 reclaim induced by memory.reclaim. 1270 1271 memory.peak 1272 A read-only single value file which exists on non-root 1273 cgroups. 1274 1275 The max memory usage recorded for the cgroup and its 1276 descendants since the creation of the cgroup. 1277 1278 memory.oom.group 1279 A read-write single value file which exists on non-root 1280 cgroups. The default value is "0". 1281 1282 Determines whether the cgroup should be treated as 1283 an indivisible workload by the OOM killer. If set, 1284 all tasks belonging to the cgroup or to its descendants 1285 (if the memory cgroup is not a leaf cgroup) are killed 1286 together or not at all. This can be used to avoid 1287 partial kills to guarantee workload integrity. 1288 1289 Tasks with the OOM protection (oom_score_adj set to -1000) 1290 are treated as an exception and are never killed. 1291 1292 If the OOM killer is invoked in a cgroup, it's not going 1293 to kill any tasks outside of this cgroup, regardless 1294 memory.oom.group values of ancestor cgroups. 1295 1296 memory.events 1297 A read-only flat-keyed file which exists on non-root cgroups. 1298 The following entries are defined. Unless specified 1299 otherwise, a value change in this file generates a file 1300 modified event. 1301 1302 Note that all fields in this file are hierarchical and the 1303 file modified event can be generated due to an event down the 1304 hierarchy. For the local events at the cgroup level see 1305 memory.events.local. 1306 1307 low 1308 The number of times the cgroup is reclaimed due to 1309 high memory pressure even though its usage is under 1310 the low boundary. This usually indicates that the low 1311 boundary is over-committed. 1312 1313 high 1314 The number of times processes of the cgroup are 1315 throttled and routed to perform direct memory reclaim 1316 because the high memory boundary was exceeded. For a 1317 cgroup whose memory usage is capped by the high limit 1318 rather than global memory pressure, this event's 1319 occurrences are expected. 1320 1321 max 1322 The number of times the cgroup's memory usage was 1323 about to go over the max boundary. If direct reclaim 1324 fails to bring it down, the cgroup goes to OOM state. 1325 1326 oom 1327 The number of time the cgroup's memory usage was 1328 reached the limit and allocation was about to fail. 1329 1330 This event is not raised if the OOM killer is not 1331 considered as an option, e.g. for failed high-order 1332 allocations or if caller asked to not retry attempts. 1333 1334 oom_kill 1335 The number of processes belonging to this cgroup 1336 killed by any kind of OOM killer. 1337 1338 oom_group_kill 1339 The number of times a group OOM has occurred. 1340 1341 memory.events.local 1342 Similar to memory.events but the fields in the file are local 1343 to the cgroup i.e. not hierarchical. The file modified event 1344 generated on this file reflects only the local events. 1345 1346 memory.stat 1347 A read-only flat-keyed file which exists on non-root cgroups. 1348 1349 This breaks down the cgroup's memory footprint into different 1350 types of memory, type-specific details, and other information 1351 on the state and past events of the memory management system. 1352 1353 All memory amounts are in bytes. 1354 1355 The entries are ordered to be human readable, and new entries 1356 can show up in the middle. Don't rely on items remaining in a 1357 fixed position; use the keys to look up specific values! 1358 1359 If the entry has no per-node counter (or not show in the 1360 memory.numa_stat). We use 'npn' (non-per-node) as the tag 1361 to indicate that it will not show in the memory.numa_stat. 1362 1363 anon 1364 Amount of memory used in anonymous mappings such as 1365 brk(), sbrk(), and mmap(MAP_ANONYMOUS) 1366 1367 file 1368 Amount of memory used to cache filesystem data, 1369 including tmpfs and shared memory. 1370 1371 kernel (npn) 1372 Amount of total kernel memory, including 1373 (kernel_stack, pagetables, percpu, vmalloc, slab) in 1374 addition to other kernel memory use cases. 1375 1376 kernel_stack 1377 Amount of memory allocated to kernel stacks. 1378 1379 pagetables 1380 Amount of memory allocated for page tables. 1381 1382 sec_pagetables 1383 Amount of memory allocated for secondary page tables, 1384 this currently includes KVM mmu allocations on x86 1385 and arm64. 1386 1387 percpu (npn) 1388 Amount of memory used for storing per-cpu kernel 1389 data structures. 1390 1391 sock (npn) 1392 Amount of memory used in network transmission buffers 1393 1394 vmalloc (npn) 1395 Amount of memory used for vmap backed memory. 1396 1397 shmem 1398 Amount of cached filesystem data that is swap-backed, 1399 such as tmpfs, shm segments, shared anonymous mmap()s 1400 1401 zswap 1402 Amount of memory consumed by the zswap compression backend. 1403 1404 zswapped 1405 Amount of application memory swapped out to zswap. 1406 1407 file_mapped 1408 Amount of cached filesystem data mapped with mmap() 1409 1410 file_dirty 1411 Amount of cached filesystem data that was modified but 1412 not yet written back to disk 1413 1414 file_writeback 1415 Amount of cached filesystem data that was modified and 1416 is currently being written back to disk 1417 1418 swapcached 1419 Amount of swap cached in memory. The swapcache is accounted 1420 against both memory and swap usage. 1421 1422 anon_thp 1423 Amount of memory used in anonymous mappings backed by 1424 transparent hugepages 1425 1426 file_thp 1427 Amount of cached filesystem data backed by transparent 1428 hugepages 1429 1430 shmem_thp 1431 Amount of shm, tmpfs, shared anonymous mmap()s backed by 1432 transparent hugepages 1433 1434 inactive_anon, active_anon, inactive_file, active_file, unevictable 1435 Amount of memory, swap-backed and filesystem-backed, 1436 on the internal memory management lists used by the 1437 page reclaim algorithm. 1438 1439 As these represent internal list state (eg. shmem pages are on anon 1440 memory management lists), inactive_foo + active_foo may not be equal to 1441 the value for the foo counter, since the foo counter is type-based, not 1442 list-based. 1443 1444 slab_reclaimable 1445 Part of "slab" that might be reclaimed, such as 1446 dentries and inodes. 1447 1448 slab_unreclaimable 1449 Part of "slab" that cannot be reclaimed on memory 1450 pressure. 1451 1452 slab (npn) 1453 Amount of memory used for storing in-kernel data 1454 structures. 1455 1456 workingset_refault_anon 1457 Number of refaults of previously evicted anonymous pages. 1458 1459 workingset_refault_file 1460 Number of refaults of previously evicted file pages. 1461 1462 workingset_activate_anon 1463 Number of refaulted anonymous pages that were immediately 1464 activated. 1465 1466 workingset_activate_file 1467 Number of refaulted file pages that were immediately activated. 1468 1469 workingset_restore_anon 1470 Number of restored anonymous pages which have been detected as 1471 an active workingset before they got reclaimed. 1472 1473 workingset_restore_file 1474 Number of restored file pages which have been detected as an 1475 active workingset before they got reclaimed. 1476 1477 workingset_nodereclaim 1478 Number of times a shadow node has been reclaimed 1479 1480 pgscan (npn) 1481 Amount of scanned pages (in an inactive LRU list) 1482 1483 pgsteal (npn) 1484 Amount of reclaimed pages 1485 1486 pgscan_kswapd (npn) 1487 Amount of scanned pages by kswapd (in an inactive LRU list) 1488 1489 pgscan_direct (npn) 1490 Amount of scanned pages directly (in an inactive LRU list) 1491 1492 pgscan_khugepaged (npn) 1493 Amount of scanned pages by khugepaged (in an inactive LRU list) 1494 1495 pgsteal_kswapd (npn) 1496 Amount of reclaimed pages by kswapd 1497 1498 pgsteal_direct (npn) 1499 Amount of reclaimed pages directly 1500 1501 pgsteal_khugepaged (npn) 1502 Amount of reclaimed pages by khugepaged 1503 1504 pgfault (npn) 1505 Total number of page faults incurred 1506 1507 pgmajfault (npn) 1508 Number of major page faults incurred 1509 1510 pgrefill (npn) 1511 Amount of scanned pages (in an active LRU list) 1512 1513 pgactivate (npn) 1514 Amount of pages moved to the active LRU list 1515 1516 pgdeactivate (npn) 1517 Amount of pages moved to the inactive LRU list 1518 1519 pglazyfree (npn) 1520 Amount of pages postponed to be freed under memory pressure 1521 1522 pglazyfreed (npn) 1523 Amount of reclaimed lazyfree pages 1524 1525 thp_fault_alloc (npn) 1526 Number of transparent hugepages which were allocated to satisfy 1527 a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE 1528 is not set. 1529 1530 thp_collapse_alloc (npn) 1531 Number of transparent hugepages which were allocated to allow 1532 collapsing an existing range of pages. This counter is not 1533 present when CONFIG_TRANSPARENT_HUGEPAGE is not set. 1534 1535 memory.numa_stat 1536 A read-only nested-keyed file which exists on non-root cgroups. 1537 1538 This breaks down the cgroup's memory footprint into different 1539 types of memory, type-specific details, and other information 1540 per node on the state of the memory management system. 1541 1542 This is useful for providing visibility into the NUMA locality 1543 information within an memcg since the pages are allowed to be 1544 allocated from any physical node. One of the use case is evaluating 1545 application performance by combining this information with the 1546 application's CPU allocation. 1547 1548 All memory amounts are in bytes. 1549 1550 The output format of memory.numa_stat is:: 1551 1552 type N0=<bytes in node 0> N1=<bytes in node 1> ... 1553 1554 The entries are ordered to be human readable, and new entries 1555 can show up in the middle. Don't rely on items remaining in a 1556 fixed position; use the keys to look up specific values! 1557 1558 The entries can refer to the memory.stat. 1559 1560 memory.swap.current 1561 A read-only single value file which exists on non-root 1562 cgroups. 1563 1564 The total amount of swap currently being used by the cgroup 1565 and its descendants. 1566 1567 memory.swap.high 1568 A read-write single value file which exists on non-root 1569 cgroups. The default is "max". 1570 1571 Swap usage throttle limit. If a cgroup's swap usage exceeds 1572 this limit, all its further allocations will be throttled to 1573 allow userspace to implement custom out-of-memory procedures. 1574 1575 This limit marks a point of no return for the cgroup. It is NOT 1576 designed to manage the amount of swapping a workload does 1577 during regular operation. Compare to memory.swap.max, which 1578 prohibits swapping past a set amount, but lets the cgroup 1579 continue unimpeded as long as other memory can be reclaimed. 1580 1581 Healthy workloads are not expected to reach this limit. 1582 1583 memory.swap.max 1584 A read-write single value file which exists on non-root 1585 cgroups. The default is "max". 1586 1587 Swap usage hard limit. If a cgroup's swap usage reaches this 1588 limit, anonymous memory of the cgroup will not be swapped out. 1589 1590 memory.swap.events 1591 A read-only flat-keyed file which exists on non-root cgroups. 1592 The following entries are defined. Unless specified 1593 otherwise, a value change in this file generates a file 1594 modified event. 1595 1596 high 1597 The number of times the cgroup's swap usage was over 1598 the high threshold. 1599 1600 max 1601 The number of times the cgroup's swap usage was about 1602 to go over the max boundary and swap allocation 1603 failed. 1604 1605 fail 1606 The number of times swap allocation failed either 1607 because of running out of swap system-wide or max 1608 limit. 1609 1610 When reduced under the current usage, the existing swap 1611 entries are reclaimed gradually and the swap usage may stay 1612 higher than the limit for an extended period of time. This 1613 reduces the impact on the workload and memory management. 1614 1615 memory.zswap.current 1616 A read-only single value file which exists on non-root 1617 cgroups. 1618 1619 The total amount of memory consumed by the zswap compression 1620 backend. 1621 1622 memory.zswap.max 1623 A read-write single value file which exists on non-root 1624 cgroups. The default is "max". 1625 1626 Zswap usage hard limit. If a cgroup's zswap pool reaches this 1627 limit, it will refuse to take any more stores before existing 1628 entries fault back in or are written out to disk. 1629 1630 memory.pressure 1631 A read-only nested-keyed file. 1632 1633 Shows pressure stall information for memory. See 1634 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1635 1636 1637Usage Guidelines 1638~~~~~~~~~~~~~~~~ 1639 1640"memory.high" is the main mechanism to control memory usage. 1641Over-committing on high limit (sum of high limits > available memory) 1642and letting global memory pressure to distribute memory according to 1643usage is a viable strategy. 1644 1645Because breach of the high limit doesn't trigger the OOM killer but 1646throttles the offending cgroup, a management agent has ample 1647opportunities to monitor and take appropriate actions such as granting 1648more memory or terminating the workload. 1649 1650Determining whether a cgroup has enough memory is not trivial as 1651memory usage doesn't indicate whether the workload can benefit from 1652more memory. For example, a workload which writes data received from 1653network to a file can use all available memory but can also operate as 1654performant with a small amount of memory. A measure of memory 1655pressure - how much the workload is being impacted due to lack of 1656memory - is necessary to determine whether a workload needs more 1657memory; unfortunately, memory pressure monitoring mechanism isn't 1658implemented yet. 1659 1660 1661Memory Ownership 1662~~~~~~~~~~~~~~~~ 1663 1664A memory area is charged to the cgroup which instantiated it and stays 1665charged to the cgroup until the area is released. Migrating a process 1666to a different cgroup doesn't move the memory usages that it 1667instantiated while in the previous cgroup to the new cgroup. 1668 1669A memory area may be used by processes belonging to different cgroups. 1670To which cgroup the area will be charged is in-deterministic; however, 1671over time, the memory area is likely to end up in a cgroup which has 1672enough memory allowance to avoid high reclaim pressure. 1673 1674If a cgroup sweeps a considerable amount of memory which is expected 1675to be accessed repeatedly by other cgroups, it may make sense to use 1676POSIX_FADV_DONTNEED to relinquish the ownership of memory areas 1677belonging to the affected files to ensure correct memory ownership. 1678 1679 1680IO 1681-- 1682 1683The "io" controller regulates the distribution of IO resources. This 1684controller implements both weight based and absolute bandwidth or IOPS 1685limit distribution; however, weight based distribution is available 1686only if cfq-iosched is in use and neither scheme is available for 1687blk-mq devices. 1688 1689 1690IO Interface Files 1691~~~~~~~~~~~~~~~~~~ 1692 1693 io.stat 1694 A read-only nested-keyed file. 1695 1696 Lines are keyed by $MAJ:$MIN device numbers and not ordered. 1697 The following nested keys are defined. 1698 1699 ====== ===================== 1700 rbytes Bytes read 1701 wbytes Bytes written 1702 rios Number of read IOs 1703 wios Number of write IOs 1704 dbytes Bytes discarded 1705 dios Number of discard IOs 1706 ====== ===================== 1707 1708 An example read output follows:: 1709 1710 8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0 1711 8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021 1712 1713 io.cost.qos 1714 A read-write nested-keyed file which exists only on the root 1715 cgroup. 1716 1717 This file configures the Quality of Service of the IO cost 1718 model based controller (CONFIG_BLK_CGROUP_IOCOST) which 1719 currently implements "io.weight" proportional control. Lines 1720 are keyed by $MAJ:$MIN device numbers and not ordered. The 1721 line for a given device is populated on the first write for 1722 the device on "io.cost.qos" or "io.cost.model". The following 1723 nested keys are defined. 1724 1725 ====== ===================================== 1726 enable Weight-based control enable 1727 ctrl "auto" or "user" 1728 rpct Read latency percentile [0, 100] 1729 rlat Read latency threshold 1730 wpct Write latency percentile [0, 100] 1731 wlat Write latency threshold 1732 min Minimum scaling percentage [1, 10000] 1733 max Maximum scaling percentage [1, 10000] 1734 ====== ===================================== 1735 1736 The controller is disabled by default and can be enabled by 1737 setting "enable" to 1. "rpct" and "wpct" parameters default 1738 to zero and the controller uses internal device saturation 1739 state to adjust the overall IO rate between "min" and "max". 1740 1741 When a better control quality is needed, latency QoS 1742 parameters can be configured. For example:: 1743 1744 8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0 1745 1746 shows that on sdb, the controller is enabled, will consider 1747 the device saturated if the 95th percentile of read completion 1748 latencies is above 75ms or write 150ms, and adjust the overall 1749 IO issue rate between 50% and 150% accordingly. 1750 1751 The lower the saturation point, the better the latency QoS at 1752 the cost of aggregate bandwidth. The narrower the allowed 1753 adjustment range between "min" and "max", the more conformant 1754 to the cost model the IO behavior. Note that the IO issue 1755 base rate may be far off from 100% and setting "min" and "max" 1756 blindly can lead to a significant loss of device capacity or 1757 control quality. "min" and "max" are useful for regulating 1758 devices which show wide temporary behavior changes - e.g. a 1759 ssd which accepts writes at the line speed for a while and 1760 then completely stalls for multiple seconds. 1761 1762 When "ctrl" is "auto", the parameters are controlled by the 1763 kernel and may change automatically. Setting "ctrl" to "user" 1764 or setting any of the percentile and latency parameters puts 1765 it into "user" mode and disables the automatic changes. The 1766 automatic mode can be restored by setting "ctrl" to "auto". 1767 1768 io.cost.model 1769 A read-write nested-keyed file which exists only on the root 1770 cgroup. 1771 1772 This file configures the cost model of the IO cost model based 1773 controller (CONFIG_BLK_CGROUP_IOCOST) which currently 1774 implements "io.weight" proportional control. Lines are keyed 1775 by $MAJ:$MIN device numbers and not ordered. The line for a 1776 given device is populated on the first write for the device on 1777 "io.cost.qos" or "io.cost.model". The following nested keys 1778 are defined. 1779 1780 ===== ================================ 1781 ctrl "auto" or "user" 1782 model The cost model in use - "linear" 1783 ===== ================================ 1784 1785 When "ctrl" is "auto", the kernel may change all parameters 1786 dynamically. When "ctrl" is set to "user" or any other 1787 parameters are written to, "ctrl" become "user" and the 1788 automatic changes are disabled. 1789 1790 When "model" is "linear", the following model parameters are 1791 defined. 1792 1793 ============= ======================================== 1794 [r|w]bps The maximum sequential IO throughput 1795 [r|w]seqiops The maximum 4k sequential IOs per second 1796 [r|w]randiops The maximum 4k random IOs per second 1797 ============= ======================================== 1798 1799 From the above, the builtin linear model determines the base 1800 costs of a sequential and random IO and the cost coefficient 1801 for the IO size. While simple, this model can cover most 1802 common device classes acceptably. 1803 1804 The IO cost model isn't expected to be accurate in absolute 1805 sense and is scaled to the device behavior dynamically. 1806 1807 If needed, tools/cgroup/iocost_coef_gen.py can be used to 1808 generate device-specific coefficients. 1809 1810 io.weight 1811 A read-write flat-keyed file which exists on non-root cgroups. 1812 The default is "default 100". 1813 1814 The first line is the default weight applied to devices 1815 without specific override. The rest are overrides keyed by 1816 $MAJ:$MIN device numbers and not ordered. The weights are in 1817 the range [1, 10000] and specifies the relative amount IO time 1818 the cgroup can use in relation to its siblings. 1819 1820 The default weight can be updated by writing either "default 1821 $WEIGHT" or simply "$WEIGHT". Overrides can be set by writing 1822 "$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default". 1823 1824 An example read output follows:: 1825 1826 default 100 1827 8:16 200 1828 8:0 50 1829 1830 io.max 1831 A read-write nested-keyed file which exists on non-root 1832 cgroups. 1833 1834 BPS and IOPS based IO limit. Lines are keyed by $MAJ:$MIN 1835 device numbers and not ordered. The following nested keys are 1836 defined. 1837 1838 ===== ================================== 1839 rbps Max read bytes per second 1840 wbps Max write bytes per second 1841 riops Max read IO operations per second 1842 wiops Max write IO operations per second 1843 ===== ================================== 1844 1845 When writing, any number of nested key-value pairs can be 1846 specified in any order. "max" can be specified as the value 1847 to remove a specific limit. If the same key is specified 1848 multiple times, the outcome is undefined. 1849 1850 BPS and IOPS are measured in each IO direction and IOs are 1851 delayed if limit is reached. Temporary bursts are allowed. 1852 1853 Setting read limit at 2M BPS and write at 120 IOPS for 8:16:: 1854 1855 echo "8:16 rbps=2097152 wiops=120" > io.max 1856 1857 Reading returns the following:: 1858 1859 8:16 rbps=2097152 wbps=max riops=max wiops=120 1860 1861 Write IOPS limit can be removed by writing the following:: 1862 1863 echo "8:16 wiops=max" > io.max 1864 1865 Reading now returns the following:: 1866 1867 8:16 rbps=2097152 wbps=max riops=max wiops=max 1868 1869 io.pressure 1870 A read-only nested-keyed file. 1871 1872 Shows pressure stall information for IO. See 1873 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1874 1875 1876Writeback 1877~~~~~~~~~ 1878 1879Page cache is dirtied through buffered writes and shared mmaps and 1880written asynchronously to the backing filesystem by the writeback 1881mechanism. Writeback sits between the memory and IO domains and 1882regulates the proportion of dirty memory by balancing dirtying and 1883write IOs. 1884 1885The io controller, in conjunction with the memory controller, 1886implements control of page cache writeback IOs. The memory controller 1887defines the memory domain that dirty memory ratio is calculated and 1888maintained for and the io controller defines the io domain which 1889writes out dirty pages for the memory domain. Both system-wide and 1890per-cgroup dirty memory states are examined and the more restrictive 1891of the two is enforced. 1892 1893cgroup writeback requires explicit support from the underlying 1894filesystem. Currently, cgroup writeback is implemented on ext2, ext4, 1895btrfs, f2fs, and xfs. On other filesystems, all writeback IOs are 1896attributed to the root cgroup. 1897 1898There are inherent differences in memory and writeback management 1899which affects how cgroup ownership is tracked. Memory is tracked per 1900page while writeback per inode. For the purpose of writeback, an 1901inode is assigned to a cgroup and all IO requests to write dirty pages 1902from the inode are attributed to that cgroup. 1903 1904As cgroup ownership for memory is tracked per page, there can be pages 1905which are associated with different cgroups than the one the inode is 1906associated with. These are called foreign pages. The writeback 1907constantly keeps track of foreign pages and, if a particular foreign 1908cgroup becomes the majority over a certain period of time, switches 1909the ownership of the inode to that cgroup. 1910 1911While this model is enough for most use cases where a given inode is 1912mostly dirtied by a single cgroup even when the main writing cgroup 1913changes over time, use cases where multiple cgroups write to a single 1914inode simultaneously are not supported well. In such circumstances, a 1915significant portion of IOs are likely to be attributed incorrectly. 1916As memory controller assigns page ownership on the first use and 1917doesn't update it until the page is released, even if writeback 1918strictly follows page ownership, multiple cgroups dirtying overlapping 1919areas wouldn't work as expected. It's recommended to avoid such usage 1920patterns. 1921 1922The sysctl knobs which affect writeback behavior are applied to cgroup 1923writeback as follows. 1924 1925 vm.dirty_background_ratio, vm.dirty_ratio 1926 These ratios apply the same to cgroup writeback with the 1927 amount of available memory capped by limits imposed by the 1928 memory controller and system-wide clean memory. 1929 1930 vm.dirty_background_bytes, vm.dirty_bytes 1931 For cgroup writeback, this is calculated into ratio against 1932 total available memory and applied the same way as 1933 vm.dirty[_background]_ratio. 1934 1935 1936IO Latency 1937~~~~~~~~~~ 1938 1939This is a cgroup v2 controller for IO workload protection. You provide a group 1940with a latency target, and if the average latency exceeds that target the 1941controller will throttle any peers that have a lower latency target than the 1942protected workload. 1943 1944The limits are only applied at the peer level in the hierarchy. This means that 1945in the diagram below, only groups A, B, and C will influence each other, and 1946groups D and F will influence each other. Group G will influence nobody:: 1947 1948 [root] 1949 / | \ 1950 A B C 1951 / \ | 1952 D F G 1953 1954 1955So the ideal way to configure this is to set io.latency in groups A, B, and C. 1956Generally you do not want to set a value lower than the latency your device 1957supports. Experiment to find the value that works best for your workload. 1958Start at higher than the expected latency for your device and watch the 1959avg_lat value in io.stat for your workload group to get an idea of the 1960latency you see during normal operation. Use the avg_lat value as a basis for 1961your real setting, setting at 10-15% higher than the value in io.stat. 1962 1963How IO Latency Throttling Works 1964~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1965 1966io.latency is work conserving; so as long as everybody is meeting their latency 1967target the controller doesn't do anything. Once a group starts missing its 1968target it begins throttling any peer group that has a higher target than itself. 1969This throttling takes 2 forms: 1970 1971- Queue depth throttling. This is the number of outstanding IO's a group is 1972 allowed to have. We will clamp down relatively quickly, starting at no limit 1973 and going all the way down to 1 IO at a time. 1974 1975- Artificial delay induction. There are certain types of IO that cannot be 1976 throttled without possibly adversely affecting higher priority groups. This 1977 includes swapping and metadata IO. These types of IO are allowed to occur 1978 normally, however they are "charged" to the originating group. If the 1979 originating group is being throttled you will see the use_delay and delay 1980 fields in io.stat increase. The delay value is how many microseconds that are 1981 being added to any process that runs in this group. Because this number can 1982 grow quite large if there is a lot of swapping or metadata IO occurring we 1983 limit the individual delay events to 1 second at a time. 1984 1985Once the victimized group starts meeting its latency target again it will start 1986unthrottling any peer groups that were throttled previously. If the victimized 1987group simply stops doing IO the global counter will unthrottle appropriately. 1988 1989IO Latency Interface Files 1990~~~~~~~~~~~~~~~~~~~~~~~~~~ 1991 1992 io.latency 1993 This takes a similar format as the other controllers. 1994 1995 "MAJOR:MINOR target=<target time in microseconds>" 1996 1997 io.stat 1998 If the controller is enabled you will see extra stats in io.stat in 1999 addition to the normal ones. 2000 2001 depth 2002 This is the current queue depth for the group. 2003 2004 avg_lat 2005 This is an exponential moving average with a decay rate of 1/exp 2006 bound by the sampling interval. The decay rate interval can be 2007 calculated by multiplying the win value in io.stat by the 2008 corresponding number of samples based on the win value. 2009 2010 win 2011 The sampling window size in milliseconds. This is the minimum 2012 duration of time between evaluation events. Windows only elapse 2013 with IO activity. Idle periods extend the most recent window. 2014 2015IO Priority 2016~~~~~~~~~~~ 2017 2018A single attribute controls the behavior of the I/O priority cgroup policy, 2019namely the blkio.prio.class attribute. The following values are accepted for 2020that attribute: 2021 2022 no-change 2023 Do not modify the I/O priority class. 2024 2025 none-to-rt 2026 For requests that do not have an I/O priority class (NONE), 2027 change the I/O priority class into RT. Do not modify 2028 the I/O priority class of other requests. 2029 2030 restrict-to-be 2031 For requests that do not have an I/O priority class or that have I/O 2032 priority class RT, change it into BE. Do not modify the I/O priority 2033 class of requests that have priority class IDLE. 2034 2035 idle 2036 Change the I/O priority class of all requests into IDLE, the lowest 2037 I/O priority class. 2038 2039The following numerical values are associated with the I/O priority policies: 2040 2041+-------------+---+ 2042| no-change | 0 | 2043+-------------+---+ 2044| none-to-rt | 1 | 2045+-------------+---+ 2046| rt-to-be | 2 | 2047+-------------+---+ 2048| all-to-idle | 3 | 2049+-------------+---+ 2050 2051The numerical value that corresponds to each I/O priority class is as follows: 2052 2053+-------------------------------+---+ 2054| IOPRIO_CLASS_NONE | 0 | 2055+-------------------------------+---+ 2056| IOPRIO_CLASS_RT (real-time) | 1 | 2057+-------------------------------+---+ 2058| IOPRIO_CLASS_BE (best effort) | 2 | 2059+-------------------------------+---+ 2060| IOPRIO_CLASS_IDLE | 3 | 2061+-------------------------------+---+ 2062 2063The algorithm to set the I/O priority class for a request is as follows: 2064 2065- Translate the I/O priority class policy into a number. 2066- Change the request I/O priority class into the maximum of the I/O priority 2067 class policy number and the numerical I/O priority class. 2068 2069PID 2070--- 2071 2072The process number controller is used to allow a cgroup to stop any 2073new tasks from being fork()'d or clone()'d after a specified limit is 2074reached. 2075 2076The number of tasks in a cgroup can be exhausted in ways which other 2077controllers cannot prevent, thus warranting its own controller. For 2078example, a fork bomb is likely to exhaust the number of tasks before 2079hitting memory restrictions. 2080 2081Note that PIDs used in this controller refer to TIDs, process IDs as 2082used by the kernel. 2083 2084 2085PID Interface Files 2086~~~~~~~~~~~~~~~~~~~ 2087 2088 pids.max 2089 A read-write single value file which exists on non-root 2090 cgroups. The default is "max". 2091 2092 Hard limit of number of processes. 2093 2094 pids.current 2095 A read-only single value file which exists on all cgroups. 2096 2097 The number of processes currently in the cgroup and its 2098 descendants. 2099 2100Organisational operations are not blocked by cgroup policies, so it is 2101possible to have pids.current > pids.max. This can be done by either 2102setting the limit to be smaller than pids.current, or attaching enough 2103processes to the cgroup such that pids.current is larger than 2104pids.max. However, it is not possible to violate a cgroup PID policy 2105through fork() or clone(). These will return -EAGAIN if the creation 2106of a new process would cause a cgroup policy to be violated. 2107 2108 2109Cpuset 2110------ 2111 2112The "cpuset" controller provides a mechanism for constraining 2113the CPU and memory node placement of tasks to only the resources 2114specified in the cpuset interface files in a task's current cgroup. 2115This is especially valuable on large NUMA systems where placing jobs 2116on properly sized subsets of the systems with careful processor and 2117memory placement to reduce cross-node memory access and contention 2118can improve overall system performance. 2119 2120The "cpuset" controller is hierarchical. That means the controller 2121cannot use CPUs or memory nodes not allowed in its parent. 2122 2123 2124Cpuset Interface Files 2125~~~~~~~~~~~~~~~~~~~~~~ 2126 2127 cpuset.cpus 2128 A read-write multiple values file which exists on non-root 2129 cpuset-enabled cgroups. 2130 2131 It lists the requested CPUs to be used by tasks within this 2132 cgroup. The actual list of CPUs to be granted, however, is 2133 subjected to constraints imposed by its parent and can differ 2134 from the requested CPUs. 2135 2136 The CPU numbers are comma-separated numbers or ranges. 2137 For example:: 2138 2139 # cat cpuset.cpus 2140 0-4,6,8-10 2141 2142 An empty value indicates that the cgroup is using the same 2143 setting as the nearest cgroup ancestor with a non-empty 2144 "cpuset.cpus" or all the available CPUs if none is found. 2145 2146 The value of "cpuset.cpus" stays constant until the next update 2147 and won't be affected by any CPU hotplug events. 2148 2149 cpuset.cpus.effective 2150 A read-only multiple values file which exists on all 2151 cpuset-enabled cgroups. 2152 2153 It lists the onlined CPUs that are actually granted to this 2154 cgroup by its parent. These CPUs are allowed to be used by 2155 tasks within the current cgroup. 2156 2157 If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows 2158 all the CPUs from the parent cgroup that can be available to 2159 be used by this cgroup. Otherwise, it should be a subset of 2160 "cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus" 2161 can be granted. In this case, it will be treated just like an 2162 empty "cpuset.cpus". 2163 2164 Its value will be affected by CPU hotplug events. 2165 2166 cpuset.mems 2167 A read-write multiple values file which exists on non-root 2168 cpuset-enabled cgroups. 2169 2170 It lists the requested memory nodes to be used by tasks within 2171 this cgroup. The actual list of memory nodes granted, however, 2172 is subjected to constraints imposed by its parent and can differ 2173 from the requested memory nodes. 2174 2175 The memory node numbers are comma-separated numbers or ranges. 2176 For example:: 2177 2178 # cat cpuset.mems 2179 0-1,3 2180 2181 An empty value indicates that the cgroup is using the same 2182 setting as the nearest cgroup ancestor with a non-empty 2183 "cpuset.mems" or all the available memory nodes if none 2184 is found. 2185 2186 The value of "cpuset.mems" stays constant until the next update 2187 and won't be affected by any memory nodes hotplug events. 2188 2189 Setting a non-empty value to "cpuset.mems" causes memory of 2190 tasks within the cgroup to be migrated to the designated nodes if 2191 they are currently using memory outside of the designated nodes. 2192 2193 There is a cost for this memory migration. The migration 2194 may not be complete and some memory pages may be left behind. 2195 So it is recommended that "cpuset.mems" should be set properly 2196 before spawning new tasks into the cpuset. Even if there is 2197 a need to change "cpuset.mems" with active tasks, it shouldn't 2198 be done frequently. 2199 2200 cpuset.mems.effective 2201 A read-only multiple values file which exists on all 2202 cpuset-enabled cgroups. 2203 2204 It lists the onlined memory nodes that are actually granted to 2205 this cgroup by its parent. These memory nodes are allowed to 2206 be used by tasks within the current cgroup. 2207 2208 If "cpuset.mems" is empty, it shows all the memory nodes from the 2209 parent cgroup that will be available to be used by this cgroup. 2210 Otherwise, it should be a subset of "cpuset.mems" unless none of 2211 the memory nodes listed in "cpuset.mems" can be granted. In this 2212 case, it will be treated just like an empty "cpuset.mems". 2213 2214 Its value will be affected by memory nodes hotplug events. 2215 2216 cpuset.cpus.partition 2217 A read-write single value file which exists on non-root 2218 cpuset-enabled cgroups. This flag is owned by the parent cgroup 2219 and is not delegatable. 2220 2221 It accepts only the following input values when written to. 2222 2223 ========== ===================================== 2224 "member" Non-root member of a partition 2225 "root" Partition root 2226 "isolated" Partition root without load balancing 2227 ========== ===================================== 2228 2229 The root cgroup is always a partition root and its state 2230 cannot be changed. All other non-root cgroups start out as 2231 "member". 2232 2233 When set to "root", the current cgroup is the root of a new 2234 partition or scheduling domain that comprises itself and all 2235 its descendants except those that are separate partition roots 2236 themselves and their descendants. 2237 2238 When set to "isolated", the CPUs in that partition root will 2239 be in an isolated state without any load balancing from the 2240 scheduler. Tasks placed in such a partition with multiple 2241 CPUs should be carefully distributed and bound to each of the 2242 individual CPUs for optimal performance. 2243 2244 The value shown in "cpuset.cpus.effective" of a partition root 2245 is the CPUs that the partition root can dedicate to a potential 2246 new child partition root. The new child subtracts available 2247 CPUs from its parent "cpuset.cpus.effective". 2248 2249 A partition root ("root" or "isolated") can be in one of the 2250 two possible states - valid or invalid. An invalid partition 2251 root is in a degraded state where some state information may 2252 be retained, but behaves more like a "member". 2253 2254 All possible state transitions among "member", "root" and 2255 "isolated" are allowed. 2256 2257 On read, the "cpuset.cpus.partition" file can show the following 2258 values. 2259 2260 ============================= ===================================== 2261 "member" Non-root member of a partition 2262 "root" Partition root 2263 "isolated" Partition root without load balancing 2264 "root invalid (<reason>)" Invalid partition root 2265 "isolated invalid (<reason>)" Invalid isolated partition root 2266 ============================= ===================================== 2267 2268 In the case of an invalid partition root, a descriptive string on 2269 why the partition is invalid is included within parentheses. 2270 2271 For a partition root to become valid, the following conditions 2272 must be met. 2273 2274 1) The "cpuset.cpus" is exclusive with its siblings , i.e. they 2275 are not shared by any of its siblings (exclusivity rule). 2276 2) The parent cgroup is a valid partition root. 2277 3) The "cpuset.cpus" is not empty and must contain at least 2278 one of the CPUs from parent's "cpuset.cpus", i.e. they overlap. 2279 4) The "cpuset.cpus.effective" cannot be empty unless there is 2280 no task associated with this partition. 2281 2282 External events like hotplug or changes to "cpuset.cpus" can 2283 cause a valid partition root to become invalid and vice versa. 2284 Note that a task cannot be moved to a cgroup with empty 2285 "cpuset.cpus.effective". 2286 2287 For a valid partition root with the sibling cpu exclusivity 2288 rule enabled, changes made to "cpuset.cpus" that violate the 2289 exclusivity rule will invalidate the partition as well as its 2290 sibling partitions with conflicting cpuset.cpus values. So 2291 care must be taking in changing "cpuset.cpus". 2292 2293 A valid non-root parent partition may distribute out all its CPUs 2294 to its child partitions when there is no task associated with it. 2295 2296 Care must be taken to change a valid partition root to 2297 "member" as all its child partitions, if present, will become 2298 invalid causing disruption to tasks running in those child 2299 partitions. These inactivated partitions could be recovered if 2300 their parent is switched back to a partition root with a proper 2301 set of "cpuset.cpus". 2302 2303 Poll and inotify events are triggered whenever the state of 2304 "cpuset.cpus.partition" changes. That includes changes caused 2305 by write to "cpuset.cpus.partition", cpu hotplug or other 2306 changes that modify the validity status of the partition. 2307 This will allow user space agents to monitor unexpected changes 2308 to "cpuset.cpus.partition" without the need to do continuous 2309 polling. 2310 2311 2312Device controller 2313----------------- 2314 2315Device controller manages access to device files. It includes both 2316creation of new device files (using mknod), and access to the 2317existing device files. 2318 2319Cgroup v2 device controller has no interface files and is implemented 2320on top of cgroup BPF. To control access to device files, a user may 2321create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach 2322them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a 2323device file, corresponding BPF programs will be executed, and depending 2324on the return value the attempt will succeed or fail with -EPERM. 2325 2326A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the 2327bpf_cgroup_dev_ctx structure, which describes the device access attempt: 2328access type (mknod/read/write) and device (type, major and minor numbers). 2329If the program returns 0, the attempt fails with -EPERM, otherwise it 2330succeeds. 2331 2332An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in 2333tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree. 2334 2335 2336RDMA 2337---- 2338 2339The "rdma" controller regulates the distribution and accounting of 2340RDMA resources. 2341 2342RDMA Interface Files 2343~~~~~~~~~~~~~~~~~~~~ 2344 2345 rdma.max 2346 A readwrite nested-keyed file that exists for all the cgroups 2347 except root that describes current configured resource limit 2348 for a RDMA/IB device. 2349 2350 Lines are keyed by device name and are not ordered. 2351 Each line contains space separated resource name and its configured 2352 limit that can be distributed. 2353 2354 The following nested keys are defined. 2355 2356 ========== ============================= 2357 hca_handle Maximum number of HCA Handles 2358 hca_object Maximum number of HCA Objects 2359 ========== ============================= 2360 2361 An example for mlx4 and ocrdma device follows:: 2362 2363 mlx4_0 hca_handle=2 hca_object=2000 2364 ocrdma1 hca_handle=3 hca_object=max 2365 2366 rdma.current 2367 A read-only file that describes current resource usage. 2368 It exists for all the cgroup except root. 2369 2370 An example for mlx4 and ocrdma device follows:: 2371 2372 mlx4_0 hca_handle=1 hca_object=20 2373 ocrdma1 hca_handle=1 hca_object=23 2374 2375HugeTLB 2376------- 2377 2378The HugeTLB controller allows to limit the HugeTLB usage per control group and 2379enforces the controller limit during page fault. 2380 2381HugeTLB Interface Files 2382~~~~~~~~~~~~~~~~~~~~~~~ 2383 2384 hugetlb.<hugepagesize>.current 2385 Show current usage for "hugepagesize" hugetlb. It exists for all 2386 the cgroup except root. 2387 2388 hugetlb.<hugepagesize>.max 2389 Set/show the hard limit of "hugepagesize" hugetlb usage. 2390 The default value is "max". It exists for all the cgroup except root. 2391 2392 hugetlb.<hugepagesize>.events 2393 A read-only flat-keyed file which exists on non-root cgroups. 2394 2395 max 2396 The number of allocation failure due to HugeTLB limit 2397 2398 hugetlb.<hugepagesize>.events.local 2399 Similar to hugetlb.<hugepagesize>.events but the fields in the file 2400 are local to the cgroup i.e. not hierarchical. The file modified event 2401 generated on this file reflects only the local events. 2402 2403 hugetlb.<hugepagesize>.numa_stat 2404 Similar to memory.numa_stat, it shows the numa information of the 2405 hugetlb pages of <hugepagesize> in this cgroup. Only active in 2406 use hugetlb pages are included. The per-node values are in bytes. 2407 2408Misc 2409---- 2410 2411The Miscellaneous cgroup provides the resource limiting and tracking 2412mechanism for the scalar resources which cannot be abstracted like the other 2413cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config 2414option. 2415 2416A resource can be added to the controller via enum misc_res_type{} in the 2417include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[] 2418in the kernel/cgroup/misc.c file. Provider of the resource must set its 2419capacity prior to using the resource by calling misc_cg_set_capacity(). 2420 2421Once a capacity is set then the resource usage can be updated using charge and 2422uncharge APIs. All of the APIs to interact with misc controller are in 2423include/linux/misc_cgroup.h. 2424 2425Misc Interface Files 2426~~~~~~~~~~~~~~~~~~~~ 2427 2428Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then: 2429 2430 misc.capacity 2431 A read-only flat-keyed file shown only in the root cgroup. It shows 2432 miscellaneous scalar resources available on the platform along with 2433 their quantities:: 2434 2435 $ cat misc.capacity 2436 res_a 50 2437 res_b 10 2438 2439 misc.current 2440 A read-only flat-keyed file shown in the non-root cgroups. It shows 2441 the current usage of the resources in the cgroup and its children.:: 2442 2443 $ cat misc.current 2444 res_a 3 2445 res_b 0 2446 2447 misc.max 2448 A read-write flat-keyed file shown in the non root cgroups. Allowed 2449 maximum usage of the resources in the cgroup and its children.:: 2450 2451 $ cat misc.max 2452 res_a max 2453 res_b 4 2454 2455 Limit can be set by:: 2456 2457 # echo res_a 1 > misc.max 2458 2459 Limit can be set to max by:: 2460 2461 # echo res_a max > misc.max 2462 2463 Limits can be set higher than the capacity value in the misc.capacity 2464 file. 2465 2466 misc.events 2467 A read-only flat-keyed file which exists on non-root cgroups. The 2468 following entries are defined. Unless specified otherwise, a value 2469 change in this file generates a file modified event. All fields in 2470 this file are hierarchical. 2471 2472 max 2473 The number of times the cgroup's resource usage was 2474 about to go over the max boundary. 2475 2476Migration and Ownership 2477~~~~~~~~~~~~~~~~~~~~~~~ 2478 2479A miscellaneous scalar resource is charged to the cgroup in which it is used 2480first, and stays charged to that cgroup until that resource is freed. Migrating 2481a process to a different cgroup does not move the charge to the destination 2482cgroup where the process has moved. 2483 2484Others 2485------ 2486 2487perf_event 2488~~~~~~~~~~ 2489 2490perf_event controller, if not mounted on a legacy hierarchy, is 2491automatically enabled on the v2 hierarchy so that perf events can 2492always be filtered by cgroup v2 path. The controller can still be 2493moved to a legacy hierarchy after v2 hierarchy is populated. 2494 2495 2496Non-normative information 2497------------------------- 2498 2499This section contains information that isn't considered to be a part of 2500the stable kernel API and so is subject to change. 2501 2502 2503CPU controller root cgroup process behaviour 2504~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2505 2506When distributing CPU cycles in the root cgroup each thread in this 2507cgroup is treated as if it was hosted in a separate child cgroup of the 2508root cgroup. This child cgroup weight is dependent on its thread nice 2509level. 2510 2511For details of this mapping see sched_prio_to_weight array in 2512kernel/sched/core.c file (values from this array should be scaled 2513appropriately so the neutral - nice 0 - value is 100 instead of 1024). 2514 2515 2516IO controller root cgroup process behaviour 2517~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2518 2519Root cgroup processes are hosted in an implicit leaf child node. 2520When distributing IO resources this implicit child node is taken into 2521account as if it was a normal child cgroup of the root cgroup with a 2522weight value of 200. 2523 2524 2525Namespace 2526========= 2527 2528Basics 2529------ 2530 2531cgroup namespace provides a mechanism to virtualize the view of the 2532"/proc/$PID/cgroup" file and cgroup mounts. The CLONE_NEWCGROUP clone 2533flag can be used with clone(2) and unshare(2) to create a new cgroup 2534namespace. The process running inside the cgroup namespace will have 2535its "/proc/$PID/cgroup" output restricted to cgroupns root. The 2536cgroupns root is the cgroup of the process at the time of creation of 2537the cgroup namespace. 2538 2539Without cgroup namespace, the "/proc/$PID/cgroup" file shows the 2540complete path of the cgroup of a process. In a container setup where 2541a set of cgroups and namespaces are intended to isolate processes the 2542"/proc/$PID/cgroup" file may leak potential system level information 2543to the isolated processes. For example:: 2544 2545 # cat /proc/self/cgroup 2546 0::/batchjobs/container_id1 2547 2548The path '/batchjobs/container_id1' can be considered as system-data 2549and undesirable to expose to the isolated processes. cgroup namespace 2550can be used to restrict visibility of this path. For example, before 2551creating a cgroup namespace, one would see:: 2552 2553 # ls -l /proc/self/ns/cgroup 2554 lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835] 2555 # cat /proc/self/cgroup 2556 0::/batchjobs/container_id1 2557 2558After unsharing a new namespace, the view changes:: 2559 2560 # ls -l /proc/self/ns/cgroup 2561 lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183] 2562 # cat /proc/self/cgroup 2563 0::/ 2564 2565When some thread from a multi-threaded process unshares its cgroup 2566namespace, the new cgroupns gets applied to the entire process (all 2567the threads). This is natural for the v2 hierarchy; however, for the 2568legacy hierarchies, this may be unexpected. 2569 2570A cgroup namespace is alive as long as there are processes inside or 2571mounts pinning it. When the last usage goes away, the cgroup 2572namespace is destroyed. The cgroupns root and the actual cgroups 2573remain. 2574 2575 2576The Root and Views 2577------------------ 2578 2579The 'cgroupns root' for a cgroup namespace is the cgroup in which the 2580process calling unshare(2) is running. For example, if a process in 2581/batchjobs/container_id1 cgroup calls unshare, cgroup 2582/batchjobs/container_id1 becomes the cgroupns root. For the 2583init_cgroup_ns, this is the real root ('/') cgroup. 2584 2585The cgroupns root cgroup does not change even if the namespace creator 2586process later moves to a different cgroup:: 2587 2588 # ~/unshare -c # unshare cgroupns in some cgroup 2589 # cat /proc/self/cgroup 2590 0::/ 2591 # mkdir sub_cgrp_1 2592 # echo 0 > sub_cgrp_1/cgroup.procs 2593 # cat /proc/self/cgroup 2594 0::/sub_cgrp_1 2595 2596Each process gets its namespace-specific view of "/proc/$PID/cgroup" 2597 2598Processes running inside the cgroup namespace will be able to see 2599cgroup paths (in /proc/self/cgroup) only inside their root cgroup. 2600From within an unshared cgroupns:: 2601 2602 # sleep 100000 & 2603 [1] 7353 2604 # echo 7353 > sub_cgrp_1/cgroup.procs 2605 # cat /proc/7353/cgroup 2606 0::/sub_cgrp_1 2607 2608From the initial cgroup namespace, the real cgroup path will be 2609visible:: 2610 2611 $ cat /proc/7353/cgroup 2612 0::/batchjobs/container_id1/sub_cgrp_1 2613 2614From a sibling cgroup namespace (that is, a namespace rooted at a 2615different cgroup), the cgroup path relative to its own cgroup 2616namespace root will be shown. For instance, if PID 7353's cgroup 2617namespace root is at '/batchjobs/container_id2', then it will see:: 2618 2619 # cat /proc/7353/cgroup 2620 0::/../container_id2/sub_cgrp_1 2621 2622Note that the relative path always starts with '/' to indicate that 2623its relative to the cgroup namespace root of the caller. 2624 2625 2626Migration and setns(2) 2627---------------------- 2628 2629Processes inside a cgroup namespace can move into and out of the 2630namespace root if they have proper access to external cgroups. For 2631example, from inside a namespace with cgroupns root at 2632/batchjobs/container_id1, and assuming that the global hierarchy is 2633still accessible inside cgroupns:: 2634 2635 # cat /proc/7353/cgroup 2636 0::/sub_cgrp_1 2637 # echo 7353 > batchjobs/container_id2/cgroup.procs 2638 # cat /proc/7353/cgroup 2639 0::/../container_id2 2640 2641Note that this kind of setup is not encouraged. A task inside cgroup 2642namespace should only be exposed to its own cgroupns hierarchy. 2643 2644setns(2) to another cgroup namespace is allowed when: 2645 2646(a) the process has CAP_SYS_ADMIN against its current user namespace 2647(b) the process has CAP_SYS_ADMIN against the target cgroup 2648 namespace's userns 2649 2650No implicit cgroup changes happen with attaching to another cgroup 2651namespace. It is expected that the someone moves the attaching 2652process under the target cgroup namespace root. 2653 2654 2655Interaction with Other Namespaces 2656--------------------------------- 2657 2658Namespace specific cgroup hierarchy can be mounted by a process 2659running inside a non-init cgroup namespace:: 2660 2661 # mount -t cgroup2 none $MOUNT_POINT 2662 2663This will mount the unified cgroup hierarchy with cgroupns root as the 2664filesystem root. The process needs CAP_SYS_ADMIN against its user and 2665mount namespaces. 2666 2667The virtualization of /proc/self/cgroup file combined with restricting 2668the view of cgroup hierarchy by namespace-private cgroupfs mount 2669provides a properly isolated cgroup view inside the container. 2670 2671 2672Information on Kernel Programming 2673================================= 2674 2675This section contains kernel programming information in the areas 2676where interacting with cgroup is necessary. cgroup core and 2677controllers are not covered. 2678 2679 2680Filesystem Support for Writeback 2681-------------------------------- 2682 2683A filesystem can support cgroup writeback by updating 2684address_space_operations->writepage[s]() to annotate bio's using the 2685following two functions. 2686 2687 wbc_init_bio(@wbc, @bio) 2688 Should be called for each bio carrying writeback data and 2689 associates the bio with the inode's owner cgroup and the 2690 corresponding request queue. This must be called after 2691 a queue (device) has been associated with the bio and 2692 before submission. 2693 2694 wbc_account_cgroup_owner(@wbc, @page, @bytes) 2695 Should be called for each data segment being written out. 2696 While this function doesn't care exactly when it's called 2697 during the writeback session, it's the easiest and most 2698 natural to call it as data segments are added to a bio. 2699 2700With writeback bio's annotated, cgroup support can be enabled per 2701super_block by setting SB_I_CGROUPWB in ->s_iflags. This allows for 2702selective disabling of cgroup writeback support which is helpful when 2703certain filesystem features, e.g. journaled data mode, are 2704incompatible. 2705 2706wbc_init_bio() binds the specified bio to its cgroup. Depending on 2707the configuration, the bio may be executed at a lower priority and if 2708the writeback session is holding shared resources, e.g. a journal 2709entry, may lead to priority inversion. There is no one easy solution 2710for the problem. Filesystems can try to work around specific problem 2711cases by skipping wbc_init_bio() and using bio_associate_blkg() 2712directly. 2713 2714 2715Deprecated v1 Core Features 2716=========================== 2717 2718- Multiple hierarchies including named ones are not supported. 2719 2720- All v1 mount options are not supported. 2721 2722- The "tasks" file is removed and "cgroup.procs" is not sorted. 2723 2724- "cgroup.clone_children" is removed. 2725 2726- /proc/cgroups is meaningless for v2. Use "cgroup.controllers" file 2727 at the root instead. 2728 2729 2730Issues with v1 and Rationales for v2 2731==================================== 2732 2733Multiple Hierarchies 2734-------------------- 2735 2736cgroup v1 allowed an arbitrary number of hierarchies and each 2737hierarchy could host any number of controllers. While this seemed to 2738provide a high level of flexibility, it wasn't useful in practice. 2739 2740For example, as there is only one instance of each controller, utility 2741type controllers such as freezer which can be useful in all 2742hierarchies could only be used in one. The issue is exacerbated by 2743the fact that controllers couldn't be moved to another hierarchy once 2744hierarchies were populated. Another issue was that all controllers 2745bound to a hierarchy were forced to have exactly the same view of the 2746hierarchy. It wasn't possible to vary the granularity depending on 2747the specific controller. 2748 2749In practice, these issues heavily limited which controllers could be 2750put on the same hierarchy and most configurations resorted to putting 2751each controller on its own hierarchy. Only closely related ones, such 2752as the cpu and cpuacct controllers, made sense to be put on the same 2753hierarchy. This often meant that userland ended up managing multiple 2754similar hierarchies repeating the same steps on each hierarchy 2755whenever a hierarchy management operation was necessary. 2756 2757Furthermore, support for multiple hierarchies came at a steep cost. 2758It greatly complicated cgroup core implementation but more importantly 2759the support for multiple hierarchies restricted how cgroup could be 2760used in general and what controllers was able to do. 2761 2762There was no limit on how many hierarchies there might be, which meant 2763that a thread's cgroup membership couldn't be described in finite 2764length. The key might contain any number of entries and was unlimited 2765in length, which made it highly awkward to manipulate and led to 2766addition of controllers which existed only to identify membership, 2767which in turn exacerbated the original problem of proliferating number 2768of hierarchies. 2769 2770Also, as a controller couldn't have any expectation regarding the 2771topologies of hierarchies other controllers might be on, each 2772controller had to assume that all other controllers were attached to 2773completely orthogonal hierarchies. This made it impossible, or at 2774least very cumbersome, for controllers to cooperate with each other. 2775 2776In most use cases, putting controllers on hierarchies which are 2777completely orthogonal to each other isn't necessary. What usually is 2778called for is the ability to have differing levels of granularity 2779depending on the specific controller. In other words, hierarchy may 2780be collapsed from leaf towards root when viewed from specific 2781controllers. For example, a given configuration might not care about 2782how memory is distributed beyond a certain level while still wanting 2783to control how CPU cycles are distributed. 2784 2785 2786Thread Granularity 2787------------------ 2788 2789cgroup v1 allowed threads of a process to belong to different cgroups. 2790This didn't make sense for some controllers and those controllers 2791ended up implementing different ways to ignore such situations but 2792much more importantly it blurred the line between API exposed to 2793individual applications and system management interface. 2794 2795Generally, in-process knowledge is available only to the process 2796itself; thus, unlike service-level organization of processes, 2797categorizing threads of a process requires active participation from 2798the application which owns the target process. 2799 2800cgroup v1 had an ambiguously defined delegation model which got abused 2801in combination with thread granularity. cgroups were delegated to 2802individual applications so that they can create and manage their own 2803sub-hierarchies and control resource distributions along them. This 2804effectively raised cgroup to the status of a syscall-like API exposed 2805to lay programs. 2806 2807First of all, cgroup has a fundamentally inadequate interface to be 2808exposed this way. For a process to access its own knobs, it has to 2809extract the path on the target hierarchy from /proc/self/cgroup, 2810construct the path by appending the name of the knob to the path, open 2811and then read and/or write to it. This is not only extremely clunky 2812and unusual but also inherently racy. There is no conventional way to 2813define transaction across the required steps and nothing can guarantee 2814that the process would actually be operating on its own sub-hierarchy. 2815 2816cgroup controllers implemented a number of knobs which would never be 2817accepted as public APIs because they were just adding control knobs to 2818system-management pseudo filesystem. cgroup ended up with interface 2819knobs which were not properly abstracted or refined and directly 2820revealed kernel internal details. These knobs got exposed to 2821individual applications through the ill-defined delegation mechanism 2822effectively abusing cgroup as a shortcut to implementing public APIs 2823without going through the required scrutiny. 2824 2825This was painful for both userland and kernel. Userland ended up with 2826misbehaving and poorly abstracted interfaces and kernel exposing and 2827locked into constructs inadvertently. 2828 2829 2830Competition Between Inner Nodes and Threads 2831------------------------------------------- 2832 2833cgroup v1 allowed threads to be in any cgroups which created an 2834interesting problem where threads belonging to a parent cgroup and its 2835children cgroups competed for resources. This was nasty as two 2836different types of entities competed and there was no obvious way to 2837settle it. Different controllers did different things. 2838 2839The cpu controller considered threads and cgroups as equivalents and 2840mapped nice levels to cgroup weights. This worked for some cases but 2841fell flat when children wanted to be allocated specific ratios of CPU 2842cycles and the number of internal threads fluctuated - the ratios 2843constantly changed as the number of competing entities fluctuated. 2844There also were other issues. The mapping from nice level to weight 2845wasn't obvious or universal, and there were various other knobs which 2846simply weren't available for threads. 2847 2848The io controller implicitly created a hidden leaf node for each 2849cgroup to host the threads. The hidden leaf had its own copies of all 2850the knobs with ``leaf_`` prefixed. While this allowed equivalent 2851control over internal threads, it was with serious drawbacks. It 2852always added an extra layer of nesting which wouldn't be necessary 2853otherwise, made the interface messy and significantly complicated the 2854implementation. 2855 2856The memory controller didn't have a way to control what happened 2857between internal tasks and child cgroups and the behavior was not 2858clearly defined. There were attempts to add ad-hoc behaviors and 2859knobs to tailor the behavior to specific workloads which would have 2860led to problems extremely difficult to resolve in the long term. 2861 2862Multiple controllers struggled with internal tasks and came up with 2863different ways to deal with it; unfortunately, all the approaches were 2864severely flawed and, furthermore, the widely different behaviors 2865made cgroup as a whole highly inconsistent. 2866 2867This clearly is a problem which needs to be addressed from cgroup core 2868in a uniform way. 2869 2870 2871Other Interface Issues 2872---------------------- 2873 2874cgroup v1 grew without oversight and developed a large number of 2875idiosyncrasies and inconsistencies. One issue on the cgroup core side 2876was how an empty cgroup was notified - a userland helper binary was 2877forked and executed for each event. The event delivery wasn't 2878recursive or delegatable. The limitations of the mechanism also led 2879to in-kernel event delivery filtering mechanism further complicating 2880the interface. 2881 2882Controller interfaces were problematic too. An extreme example is 2883controllers completely ignoring hierarchical organization and treating 2884all cgroups as if they were all located directly under the root 2885cgroup. Some controllers exposed a large amount of inconsistent 2886implementation details to userland. 2887 2888There also was no consistency across controllers. When a new cgroup 2889was created, some controllers defaulted to not imposing extra 2890restrictions while others disallowed any resource usage until 2891explicitly configured. Configuration knobs for the same type of 2892control used widely differing naming schemes and formats. Statistics 2893and information knobs were named arbitrarily and used different 2894formats and units even in the same controller. 2895 2896cgroup v2 establishes common conventions where appropriate and updates 2897controllers so that they expose minimal and consistent interfaces. 2898 2899 2900Controller Issues and Remedies 2901------------------------------ 2902 2903Memory 2904~~~~~~ 2905 2906The original lower boundary, the soft limit, is defined as a limit 2907that is per default unset. As a result, the set of cgroups that 2908global reclaim prefers is opt-in, rather than opt-out. The costs for 2909optimizing these mostly negative lookups are so high that the 2910implementation, despite its enormous size, does not even provide the 2911basic desirable behavior. First off, the soft limit has no 2912hierarchical meaning. All configured groups are organized in a global 2913rbtree and treated like equal peers, regardless where they are located 2914in the hierarchy. This makes subtree delegation impossible. Second, 2915the soft limit reclaim pass is so aggressive that it not just 2916introduces high allocation latencies into the system, but also impacts 2917system performance due to overreclaim, to the point where the feature 2918becomes self-defeating. 2919 2920The memory.low boundary on the other hand is a top-down allocated 2921reserve. A cgroup enjoys reclaim protection when it's within its 2922effective low, which makes delegation of subtrees possible. It also 2923enjoys having reclaim pressure proportional to its overage when 2924above its effective low. 2925 2926The original high boundary, the hard limit, is defined as a strict 2927limit that can not budge, even if the OOM killer has to be called. 2928But this generally goes against the goal of making the most out of the 2929available memory. The memory consumption of workloads varies during 2930runtime, and that requires users to overcommit. But doing that with a 2931strict upper limit requires either a fairly accurate prediction of the 2932working set size or adding slack to the limit. Since working set size 2933estimation is hard and error prone, and getting it wrong results in 2934OOM kills, most users tend to err on the side of a looser limit and 2935end up wasting precious resources. 2936 2937The memory.high boundary on the other hand can be set much more 2938conservatively. When hit, it throttles allocations by forcing them 2939into direct reclaim to work off the excess, but it never invokes the 2940OOM killer. As a result, a high boundary that is chosen too 2941aggressively will not terminate the processes, but instead it will 2942lead to gradual performance degradation. The user can monitor this 2943and make corrections until the minimal memory footprint that still 2944gives acceptable performance is found. 2945 2946In extreme cases, with many concurrent allocations and a complete 2947breakdown of reclaim progress within the group, the high boundary can 2948be exceeded. But even then it's mostly better to satisfy the 2949allocation from the slack available in other groups or the rest of the 2950system than killing the group. Otherwise, memory.max is there to 2951limit this type of spillover and ultimately contain buggy or even 2952malicious applications. 2953 2954Setting the original memory.limit_in_bytes below the current usage was 2955subject to a race condition, where concurrent charges could cause the 2956limit setting to fail. memory.max on the other hand will first set the 2957limit to prevent new charges, and then reclaim and OOM kill until the 2958new limit is met - or the task writing to memory.max is killed. 2959 2960The combined memory+swap accounting and limiting is replaced by real 2961control over swap space. 2962 2963The main argument for a combined memory+swap facility in the original 2964cgroup design was that global or parental pressure would always be 2965able to swap all anonymous memory of a child group, regardless of the 2966child's own (possibly untrusted) configuration. However, untrusted 2967groups can sabotage swapping by other means - such as referencing its 2968anonymous memory in a tight loop - and an admin can not assume full 2969swappability when overcommitting untrusted jobs. 2970 2971For trusted jobs, on the other hand, a combined counter is not an 2972intuitive userspace interface, and it flies in the face of the idea 2973that cgroup controllers should account and limit specific physical 2974resources. Swap space is a resource like all others in the system, 2975and that's why unified hierarchy allows distributing it separately. 2976