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 memory_hugetlb_accounting 214 Count HugeTLB memory usage towards the cgroup's overall 215 memory usage for the memory controller (for the purpose of 216 statistics reporting and memory protetion). This is a new 217 behavior that could regress existing setups, so it must be 218 explicitly opted in with this mount option. 219 220 A few caveats to keep in mind: 221 222 * There is no HugeTLB pool management involved in the memory 223 controller. The pre-allocated pool does not belong to anyone. 224 Specifically, when a new HugeTLB folio is allocated to 225 the pool, it is not accounted for from the perspective of the 226 memory controller. It is only charged to a cgroup when it is 227 actually used (for e.g at page fault time). Host memory 228 overcommit management has to consider this when configuring 229 hard limits. In general, HugeTLB pool management should be 230 done via other mechanisms (such as the HugeTLB controller). 231 * Failure to charge a HugeTLB folio to the memory controller 232 results in SIGBUS. This could happen even if the HugeTLB pool 233 still has pages available (but the cgroup limit is hit and 234 reclaim attempt fails). 235 * Charging HugeTLB memory towards the memory controller affects 236 memory protection and reclaim dynamics. Any userspace tuning 237 (of low, min limits for e.g) needs to take this into account. 238 * HugeTLB pages utilized while this option is not selected 239 will not be tracked by the memory controller (even if cgroup 240 v2 is remounted later on). 241 242 243Organizing Processes and Threads 244-------------------------------- 245 246Processes 247~~~~~~~~~ 248 249Initially, only the root cgroup exists to which all processes belong. 250A child cgroup can be created by creating a sub-directory:: 251 252 # mkdir $CGROUP_NAME 253 254A given cgroup may have multiple child cgroups forming a tree 255structure. Each cgroup has a read-writable interface file 256"cgroup.procs". When read, it lists the PIDs of all processes which 257belong to the cgroup one-per-line. The PIDs are not ordered and the 258same PID may show up more than once if the process got moved to 259another cgroup and then back or the PID got recycled while reading. 260 261A process can be migrated into a cgroup by writing its PID to the 262target cgroup's "cgroup.procs" file. Only one process can be migrated 263on a single write(2) call. If a process is composed of multiple 264threads, writing the PID of any thread migrates all threads of the 265process. 266 267When a process forks a child process, the new process is born into the 268cgroup that the forking process belongs to at the time of the 269operation. After exit, a process stays associated with the cgroup 270that it belonged to at the time of exit until it's reaped; however, a 271zombie process does not appear in "cgroup.procs" and thus can't be 272moved to another cgroup. 273 274A cgroup which doesn't have any children or live processes can be 275destroyed by removing the directory. Note that a cgroup which doesn't 276have any children and is associated only with zombie processes is 277considered empty and can be removed:: 278 279 # rmdir $CGROUP_NAME 280 281"/proc/$PID/cgroup" lists a process's cgroup membership. If legacy 282cgroup is in use in the system, this file may contain multiple lines, 283one for each hierarchy. The entry for cgroup v2 is always in the 284format "0::$PATH":: 285 286 # cat /proc/842/cgroup 287 ... 288 0::/test-cgroup/test-cgroup-nested 289 290If the process becomes a zombie and the cgroup it was associated with 291is removed subsequently, " (deleted)" is appended to the path:: 292 293 # cat /proc/842/cgroup 294 ... 295 0::/test-cgroup/test-cgroup-nested (deleted) 296 297 298Threads 299~~~~~~~ 300 301cgroup v2 supports thread granularity for a subset of controllers to 302support use cases requiring hierarchical resource distribution across 303the threads of a group of processes. By default, all threads of a 304process belong to the same cgroup, which also serves as the resource 305domain to host resource consumptions which are not specific to a 306process or thread. The thread mode allows threads to be spread across 307a subtree while still maintaining the common resource domain for them. 308 309Controllers which support thread mode are called threaded controllers. 310The ones which don't are called domain controllers. 311 312Marking a cgroup threaded makes it join the resource domain of its 313parent as a threaded cgroup. The parent may be another threaded 314cgroup whose resource domain is further up in the hierarchy. The root 315of a threaded subtree, that is, the nearest ancestor which is not 316threaded, is called threaded domain or thread root interchangeably and 317serves as the resource domain for the entire subtree. 318 319Inside a threaded subtree, threads of a process can be put in 320different cgroups and are not subject to the no internal process 321constraint - threaded controllers can be enabled on non-leaf cgroups 322whether they have threads in them or not. 323 324As the threaded domain cgroup hosts all the domain resource 325consumptions of the subtree, it is considered to have internal 326resource consumptions whether there are processes in it or not and 327can't have populated child cgroups which aren't threaded. Because the 328root cgroup is not subject to no internal process constraint, it can 329serve both as a threaded domain and a parent to domain cgroups. 330 331The current operation mode or type of the cgroup is shown in the 332"cgroup.type" file which indicates whether the cgroup is a normal 333domain, a domain which is serving as the domain of a threaded subtree, 334or a threaded cgroup. 335 336On creation, a cgroup is always a domain cgroup and can be made 337threaded by writing "threaded" to the "cgroup.type" file. The 338operation is single direction:: 339 340 # echo threaded > cgroup.type 341 342Once threaded, the cgroup can't be made a domain again. To enable the 343thread mode, the following conditions must be met. 344 345- As the cgroup will join the parent's resource domain. The parent 346 must either be a valid (threaded) domain or a threaded cgroup. 347 348- When the parent is an unthreaded domain, it must not have any domain 349 controllers enabled or populated domain children. The root is 350 exempt from this requirement. 351 352Topology-wise, a cgroup can be in an invalid state. Please consider 353the following topology:: 354 355 A (threaded domain) - B (threaded) - C (domain, just created) 356 357C is created as a domain but isn't connected to a parent which can 358host child domains. C can't be used until it is turned into a 359threaded cgroup. "cgroup.type" file will report "domain (invalid)" in 360these cases. Operations which fail due to invalid topology use 361EOPNOTSUPP as the errno. 362 363A domain cgroup is turned into a threaded domain when one of its child 364cgroup becomes threaded or threaded controllers are enabled in the 365"cgroup.subtree_control" file while there are processes in the cgroup. 366A threaded domain reverts to a normal domain when the conditions 367clear. 368 369When read, "cgroup.threads" contains the list of the thread IDs of all 370threads in the cgroup. Except that the operations are per-thread 371instead of per-process, "cgroup.threads" has the same format and 372behaves the same way as "cgroup.procs". While "cgroup.threads" can be 373written to in any cgroup, as it can only move threads inside the same 374threaded domain, its operations are confined inside each threaded 375subtree. 376 377The threaded domain cgroup serves as the resource domain for the whole 378subtree, and, while the threads can be scattered across the subtree, 379all the processes are considered to be in the threaded domain cgroup. 380"cgroup.procs" in a threaded domain cgroup contains the PIDs of all 381processes in the subtree and is not readable in the subtree proper. 382However, "cgroup.procs" can be written to from anywhere in the subtree 383to migrate all threads of the matching process to the cgroup. 384 385Only threaded controllers can be enabled in a threaded subtree. When 386a threaded controller is enabled inside a threaded subtree, it only 387accounts for and controls resource consumptions associated with the 388threads in the cgroup and its descendants. All consumptions which 389aren't tied to a specific thread belong to the threaded domain cgroup. 390 391Because a threaded subtree is exempt from no internal process 392constraint, a threaded controller must be able to handle competition 393between threads in a non-leaf cgroup and its child cgroups. Each 394threaded controller defines how such competitions are handled. 395 396Currently, the following controllers are threaded and can be enabled 397in a threaded cgroup:: 398 399- cpu 400- cpuset 401- perf_event 402- pids 403 404[Un]populated Notification 405-------------------------- 406 407Each non-root cgroup has a "cgroup.events" file which contains 408"populated" field indicating whether the cgroup's sub-hierarchy has 409live processes in it. Its value is 0 if there is no live process in 410the cgroup and its descendants; otherwise, 1. poll and [id]notify 411events are triggered when the value changes. This can be used, for 412example, to start a clean-up operation after all processes of a given 413sub-hierarchy have exited. The populated state updates and 414notifications are recursive. Consider the following sub-hierarchy 415where the numbers in the parentheses represent the numbers of processes 416in each cgroup:: 417 418 A(4) - B(0) - C(1) 419 \ D(0) 420 421A, B and C's "populated" fields would be 1 while D's 0. After the one 422process in C exits, B and C's "populated" fields would flip to "0" and 423file modified events will be generated on the "cgroup.events" files of 424both cgroups. 425 426 427Controlling Controllers 428----------------------- 429 430Enabling and Disabling 431~~~~~~~~~~~~~~~~~~~~~~ 432 433Each cgroup has a "cgroup.controllers" file which lists all 434controllers available for the cgroup to enable:: 435 436 # cat cgroup.controllers 437 cpu io memory 438 439No controller is enabled by default. Controllers can be enabled and 440disabled by writing to the "cgroup.subtree_control" file:: 441 442 # echo "+cpu +memory -io" > cgroup.subtree_control 443 444Only controllers which are listed in "cgroup.controllers" can be 445enabled. When multiple operations are specified as above, either they 446all succeed or fail. If multiple operations on the same controller 447are specified, the last one is effective. 448 449Enabling a controller in a cgroup indicates that the distribution of 450the target resource across its immediate children will be controlled. 451Consider the following sub-hierarchy. The enabled controllers are 452listed in parentheses:: 453 454 A(cpu,memory) - B(memory) - C() 455 \ D() 456 457As A has "cpu" and "memory" enabled, A will control the distribution 458of CPU cycles and memory to its children, in this case, B. As B has 459"memory" enabled but not "CPU", C and D will compete freely on CPU 460cycles but their division of memory available to B will be controlled. 461 462As a controller regulates the distribution of the target resource to 463the cgroup's children, enabling it creates the controller's interface 464files in the child cgroups. In the above example, enabling "cpu" on B 465would create the "cpu." prefixed controller interface files in C and 466D. Likewise, disabling "memory" from B would remove the "memory." 467prefixed controller interface files from C and D. This means that the 468controller interface files - anything which doesn't start with 469"cgroup." are owned by the parent rather than the cgroup itself. 470 471 472Top-down Constraint 473~~~~~~~~~~~~~~~~~~~ 474 475Resources are distributed top-down and a cgroup can further distribute 476a resource only if the resource has been distributed to it from the 477parent. This means that all non-root "cgroup.subtree_control" files 478can only contain controllers which are enabled in the parent's 479"cgroup.subtree_control" file. A controller can be enabled only if 480the parent has the controller enabled and a controller can't be 481disabled if one or more children have it enabled. 482 483 484No Internal Process Constraint 485~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 486 487Non-root cgroups can distribute domain resources to their children 488only when they don't have any processes of their own. In other words, 489only domain cgroups which don't contain any processes can have domain 490controllers enabled in their "cgroup.subtree_control" files. 491 492This guarantees that, when a domain controller is looking at the part 493of the hierarchy which has it enabled, processes are always only on 494the leaves. This rules out situations where child cgroups compete 495against internal processes of the parent. 496 497The root cgroup is exempt from this restriction. Root contains 498processes and anonymous resource consumption which can't be associated 499with any other cgroups and requires special treatment from most 500controllers. How resource consumption in the root cgroup is governed 501is up to each controller (for more information on this topic please 502refer to the Non-normative information section in the Controllers 503chapter). 504 505Note that the restriction doesn't get in the way if there is no 506enabled controller in the cgroup's "cgroup.subtree_control". This is 507important as otherwise it wouldn't be possible to create children of a 508populated cgroup. To control resource distribution of a cgroup, the 509cgroup must create children and transfer all its processes to the 510children before enabling controllers in its "cgroup.subtree_control" 511file. 512 513 514Delegation 515---------- 516 517Model of Delegation 518~~~~~~~~~~~~~~~~~~~ 519 520A cgroup can be delegated in two ways. First, to a less privileged 521user by granting write access of the directory and its "cgroup.procs", 522"cgroup.threads" and "cgroup.subtree_control" files to the user. 523Second, if the "nsdelegate" mount option is set, automatically to a 524cgroup namespace on namespace creation. 525 526Because the resource control interface files in a given directory 527control the distribution of the parent's resources, the delegatee 528shouldn't be allowed to write to them. For the first method, this is 529achieved by not granting access to these files. For the second, the 530kernel rejects writes to all files other than "cgroup.procs" and 531"cgroup.subtree_control" on a namespace root from inside the 532namespace. 533 534The end results are equivalent for both delegation types. Once 535delegated, the user can build sub-hierarchy under the directory, 536organize processes inside it as it sees fit and further distribute the 537resources it received from the parent. The limits and other settings 538of all resource controllers are hierarchical and regardless of what 539happens in the delegated sub-hierarchy, nothing can escape the 540resource restrictions imposed by the parent. 541 542Currently, cgroup doesn't impose any restrictions on the number of 543cgroups in or nesting depth of a delegated sub-hierarchy; however, 544this may be limited explicitly in the future. 545 546 547Delegation Containment 548~~~~~~~~~~~~~~~~~~~~~~ 549 550A delegated sub-hierarchy is contained in the sense that processes 551can't be moved into or out of the sub-hierarchy by the delegatee. 552 553For delegations to a less privileged user, this is achieved by 554requiring the following conditions for a process with a non-root euid 555to migrate a target process into a cgroup by writing its PID to the 556"cgroup.procs" file. 557 558- The writer must have write access to the "cgroup.procs" file. 559 560- The writer must have write access to the "cgroup.procs" file of the 561 common ancestor of the source and destination cgroups. 562 563The above two constraints ensure that while a delegatee may migrate 564processes around freely in the delegated sub-hierarchy it can't pull 565in from or push out to outside the sub-hierarchy. 566 567For an example, let's assume cgroups C0 and C1 have been delegated to 568user U0 who created C00, C01 under C0 and C10 under C1 as follows and 569all processes under C0 and C1 belong to U0:: 570 571 ~~~~~~~~~~~~~ - C0 - C00 572 ~ cgroup ~ \ C01 573 ~ hierarchy ~ 574 ~~~~~~~~~~~~~ - C1 - C10 575 576Let's also say U0 wants to write the PID of a process which is 577currently in C10 into "C00/cgroup.procs". U0 has write access to the 578file; however, the common ancestor of the source cgroup C10 and the 579destination cgroup C00 is above the points of delegation and U0 would 580not have write access to its "cgroup.procs" files and thus the write 581will be denied with -EACCES. 582 583For delegations to namespaces, containment is achieved by requiring 584that both the source and destination cgroups are reachable from the 585namespace of the process which is attempting the migration. If either 586is not reachable, the migration is rejected with -ENOENT. 587 588 589Guidelines 590---------- 591 592Organize Once and Control 593~~~~~~~~~~~~~~~~~~~~~~~~~ 594 595Migrating a process across cgroups is a relatively expensive operation 596and stateful resources such as memory are not moved together with the 597process. This is an explicit design decision as there often exist 598inherent trade-offs between migration and various hot paths in terms 599of synchronization cost. 600 601As such, migrating processes across cgroups frequently as a means to 602apply different resource restrictions is discouraged. A workload 603should be assigned to a cgroup according to the system's logical and 604resource structure once on start-up. Dynamic adjustments to resource 605distribution can be made by changing controller configuration through 606the interface files. 607 608 609Avoid Name Collisions 610~~~~~~~~~~~~~~~~~~~~~ 611 612Interface files for a cgroup and its children cgroups occupy the same 613directory and it is possible to create children cgroups which collide 614with interface files. 615 616All cgroup core interface files are prefixed with "cgroup." and each 617controller's interface files are prefixed with the controller name and 618a dot. A controller's name is composed of lower case alphabets and 619'_'s but never begins with an '_' so it can be used as the prefix 620character for collision avoidance. Also, interface file names won't 621start or end with terms which are often used in categorizing workloads 622such as job, service, slice, unit or workload. 623 624cgroup doesn't do anything to prevent name collisions and it's the 625user's responsibility to avoid them. 626 627 628Resource Distribution Models 629============================ 630 631cgroup controllers implement several resource distribution schemes 632depending on the resource type and expected use cases. This section 633describes major schemes in use along with their expected behaviors. 634 635 636Weights 637------- 638 639A parent's resource is distributed by adding up the weights of all 640active children and giving each the fraction matching the ratio of its 641weight against the sum. As only children which can make use of the 642resource at the moment participate in the distribution, this is 643work-conserving. Due to the dynamic nature, this model is usually 644used for stateless resources. 645 646All weights are in the range [1, 10000] with the default at 100. This 647allows symmetric multiplicative biases in both directions at fine 648enough granularity while staying in the intuitive range. 649 650As long as the weight is in range, all configuration combinations are 651valid and there is no reason to reject configuration changes or 652process migrations. 653 654"cpu.weight" proportionally distributes CPU cycles to active children 655and is an example of this type. 656 657 658.. _cgroupv2-limits-distributor: 659 660Limits 661------ 662 663A child can only consume up to the configured amount of the resource. 664Limits can be over-committed - the sum of the limits of children can 665exceed the amount of resource available to the parent. 666 667Limits are in the range [0, max] and defaults to "max", which is noop. 668 669As limits can be over-committed, all configuration combinations are 670valid and there is no reason to reject configuration changes or 671process migrations. 672 673"io.max" limits the maximum BPS and/or IOPS that a cgroup can consume 674on an IO device and is an example of this type. 675 676.. _cgroupv2-protections-distributor: 677 678Protections 679----------- 680 681A cgroup is protected up to the configured amount of the resource 682as long as the usages of all its ancestors are under their 683protected levels. Protections can be hard guarantees or best effort 684soft boundaries. Protections can also be over-committed in which case 685only up to the amount available to the parent is protected among 686children. 687 688Protections are in the range [0, max] and defaults to 0, which is 689noop. 690 691As protections can be over-committed, all configuration combinations 692are valid and there is no reason to reject configuration changes or 693process migrations. 694 695"memory.low" implements best-effort memory protection and is an 696example of this type. 697 698 699Allocations 700----------- 701 702A cgroup is exclusively allocated a certain amount of a finite 703resource. Allocations can't be over-committed - the sum of the 704allocations of children can not exceed the amount of resource 705available to the parent. 706 707Allocations are in the range [0, max] and defaults to 0, which is no 708resource. 709 710As allocations can't be over-committed, some configuration 711combinations are invalid and should be rejected. Also, if the 712resource is mandatory for execution of processes, process migrations 713may be rejected. 714 715"cpu.rt.max" hard-allocates realtime slices and is an example of this 716type. 717 718 719Interface Files 720=============== 721 722Format 723------ 724 725All interface files should be in one of the following formats whenever 726possible:: 727 728 New-line separated values 729 (when only one value can be written at once) 730 731 VAL0\n 732 VAL1\n 733 ... 734 735 Space separated values 736 (when read-only or multiple values can be written at once) 737 738 VAL0 VAL1 ...\n 739 740 Flat keyed 741 742 KEY0 VAL0\n 743 KEY1 VAL1\n 744 ... 745 746 Nested keyed 747 748 KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01... 749 KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11... 750 ... 751 752For a writable file, the format for writing should generally match 753reading; however, controllers may allow omitting later fields or 754implement restricted shortcuts for most common use cases. 755 756For both flat and nested keyed files, only the values for a single key 757can be written at a time. For nested keyed files, the sub key pairs 758may be specified in any order and not all pairs have to be specified. 759 760 761Conventions 762----------- 763 764- Settings for a single feature should be contained in a single file. 765 766- The root cgroup should be exempt from resource control and thus 767 shouldn't have resource control interface files. 768 769- The default time unit is microseconds. If a different unit is ever 770 used, an explicit unit suffix must be present. 771 772- A parts-per quantity should use a percentage decimal with at least 773 two digit fractional part - e.g. 13.40. 774 775- If a controller implements weight based resource distribution, its 776 interface file should be named "weight" and have the range [1, 777 10000] with 100 as the default. The values are chosen to allow 778 enough and symmetric bias in both directions while keeping it 779 intuitive (the default is 100%). 780 781- If a controller implements an absolute resource guarantee and/or 782 limit, the interface files should be named "min" and "max" 783 respectively. If a controller implements best effort resource 784 guarantee and/or limit, the interface files should be named "low" 785 and "high" respectively. 786 787 In the above four control files, the special token "max" should be 788 used to represent upward infinity for both reading and writing. 789 790- If a setting has a configurable default value and keyed specific 791 overrides, the default entry should be keyed with "default" and 792 appear as the first entry in the file. 793 794 The default value can be updated by writing either "default $VAL" or 795 "$VAL". 796 797 When writing to update a specific override, "default" can be used as 798 the value to indicate removal of the override. Override entries 799 with "default" as the value must not appear when read. 800 801 For example, a setting which is keyed by major:minor device numbers 802 with integer values may look like the following:: 803 804 # cat cgroup-example-interface-file 805 default 150 806 8:0 300 807 808 The default value can be updated by:: 809 810 # echo 125 > cgroup-example-interface-file 811 812 or:: 813 814 # echo "default 125" > cgroup-example-interface-file 815 816 An override can be set by:: 817 818 # echo "8:16 170" > cgroup-example-interface-file 819 820 and cleared by:: 821 822 # echo "8:0 default" > cgroup-example-interface-file 823 # cat cgroup-example-interface-file 824 default 125 825 8:16 170 826 827- For events which are not very high frequency, an interface file 828 "events" should be created which lists event key value pairs. 829 Whenever a notifiable event happens, file modified event should be 830 generated on the file. 831 832 833Core Interface Files 834-------------------- 835 836All cgroup core files are prefixed with "cgroup." 837 838 cgroup.type 839 A read-write single value file which exists on non-root 840 cgroups. 841 842 When read, it indicates the current type of the cgroup, which 843 can be one of the following values. 844 845 - "domain" : A normal valid domain cgroup. 846 847 - "domain threaded" : A threaded domain cgroup which is 848 serving as the root of a threaded subtree. 849 850 - "domain invalid" : A cgroup which is in an invalid state. 851 It can't be populated or have controllers enabled. It may 852 be allowed to become a threaded cgroup. 853 854 - "threaded" : A threaded cgroup which is a member of a 855 threaded subtree. 856 857 A cgroup can be turned into a threaded cgroup by writing 858 "threaded" to this file. 859 860 cgroup.procs 861 A read-write new-line separated values file which exists on 862 all cgroups. 863 864 When read, it lists the PIDs of all processes which belong to 865 the cgroup one-per-line. The PIDs are not ordered and the 866 same PID may show up more than once if the process got moved 867 to another cgroup and then back or the PID got recycled while 868 reading. 869 870 A PID can be written to migrate the process associated with 871 the PID to the cgroup. The writer should match all of the 872 following conditions. 873 874 - It must have write access to the "cgroup.procs" file. 875 876 - It must have write access to the "cgroup.procs" file of the 877 common ancestor of the source and destination cgroups. 878 879 When delegating a sub-hierarchy, write access to this file 880 should be granted along with the containing directory. 881 882 In a threaded cgroup, reading this file fails with EOPNOTSUPP 883 as all the processes belong to the thread root. Writing is 884 supported and moves every thread of the process to the cgroup. 885 886 cgroup.threads 887 A read-write new-line separated values file which exists on 888 all cgroups. 889 890 When read, it lists the TIDs of all threads which belong to 891 the cgroup one-per-line. The TIDs are not ordered and the 892 same TID may show up more than once if the thread got moved to 893 another cgroup and then back or the TID got recycled while 894 reading. 895 896 A TID can be written to migrate the thread associated with the 897 TID to the cgroup. The writer should match all of the 898 following conditions. 899 900 - It must have write access to the "cgroup.threads" file. 901 902 - The cgroup that the thread is currently in must be in the 903 same resource domain as the destination cgroup. 904 905 - It must have write access to the "cgroup.procs" file of the 906 common ancestor of the source and destination cgroups. 907 908 When delegating a sub-hierarchy, write access to this file 909 should be granted along with the containing directory. 910 911 cgroup.controllers 912 A read-only space separated values file which exists on all 913 cgroups. 914 915 It shows space separated list of all controllers available to 916 the cgroup. The controllers are not ordered. 917 918 cgroup.subtree_control 919 A read-write space separated values file which exists on all 920 cgroups. Starts out empty. 921 922 When read, it shows space separated list of the controllers 923 which are enabled to control resource distribution from the 924 cgroup to its children. 925 926 Space separated list of controllers prefixed with '+' or '-' 927 can be written to enable or disable controllers. A controller 928 name prefixed with '+' enables the controller and '-' 929 disables. If a controller appears more than once on the list, 930 the last one is effective. When multiple enable and disable 931 operations are specified, either all succeed or all fail. 932 933 cgroup.events 934 A read-only flat-keyed file which exists on non-root cgroups. 935 The following entries are defined. Unless specified 936 otherwise, a value change in this file generates a file 937 modified event. 938 939 populated 940 1 if the cgroup or its descendants contains any live 941 processes; otherwise, 0. 942 frozen 943 1 if the cgroup is frozen; otherwise, 0. 944 945 cgroup.max.descendants 946 A read-write single value files. The default is "max". 947 948 Maximum allowed number of descent cgroups. 949 If the actual number of descendants is equal or larger, 950 an attempt to create a new cgroup in the hierarchy will fail. 951 952 cgroup.max.depth 953 A read-write single value files. The default is "max". 954 955 Maximum allowed descent depth below the current cgroup. 956 If the actual descent depth is equal or larger, 957 an attempt to create a new child cgroup will fail. 958 959 cgroup.stat 960 A read-only flat-keyed file with the following entries: 961 962 nr_descendants 963 Total number of visible descendant cgroups. 964 965 nr_dying_descendants 966 Total number of dying descendant cgroups. A cgroup becomes 967 dying after being deleted by a user. The cgroup will remain 968 in dying state for some time undefined time (which can depend 969 on system load) before being completely destroyed. 970 971 A process can't enter a dying cgroup under any circumstances, 972 a dying cgroup can't revive. 973 974 A dying cgroup can consume system resources not exceeding 975 limits, which were active at the moment of cgroup deletion. 976 977 cgroup.freeze 978 A read-write single value file which exists on non-root cgroups. 979 Allowed values are "0" and "1". The default is "0". 980 981 Writing "1" to the file causes freezing of the cgroup and all 982 descendant cgroups. This means that all belonging processes will 983 be stopped and will not run until the cgroup will be explicitly 984 unfrozen. Freezing of the cgroup may take some time; when this action 985 is completed, the "frozen" value in the cgroup.events control file 986 will be updated to "1" and the corresponding notification will be 987 issued. 988 989 A cgroup can be frozen either by its own settings, or by settings 990 of any ancestor cgroups. If any of ancestor cgroups is frozen, the 991 cgroup will remain frozen. 992 993 Processes in the frozen cgroup can be killed by a fatal signal. 994 They also can enter and leave a frozen cgroup: either by an explicit 995 move by a user, or if freezing of the cgroup races with fork(). 996 If a process is moved to a frozen cgroup, it stops. If a process is 997 moved out of a frozen cgroup, it becomes running. 998 999 Frozen status of a cgroup doesn't affect any cgroup tree operations: 1000 it's possible to delete a frozen (and empty) cgroup, as well as 1001 create new sub-cgroups. 1002 1003 cgroup.kill 1004 A write-only single value file which exists in non-root cgroups. 1005 The only allowed value is "1". 1006 1007 Writing "1" to the file causes the cgroup and all descendant cgroups to 1008 be killed. This means that all processes located in the affected cgroup 1009 tree will be killed via SIGKILL. 1010 1011 Killing a cgroup tree will deal with concurrent forks appropriately and 1012 is protected against migrations. 1013 1014 In a threaded cgroup, writing this file fails with EOPNOTSUPP as 1015 killing cgroups is a process directed operation, i.e. it affects 1016 the whole thread-group. 1017 1018 cgroup.pressure 1019 A read-write single value file that allowed values are "0" and "1". 1020 The default is "1". 1021 1022 Writing "0" to the file will disable the cgroup PSI accounting. 1023 Writing "1" to the file will re-enable the cgroup PSI accounting. 1024 1025 This control attribute is not hierarchical, so disable or enable PSI 1026 accounting in a cgroup does not affect PSI accounting in descendants 1027 and doesn't need pass enablement via ancestors from root. 1028 1029 The reason this control attribute exists is that PSI accounts stalls for 1030 each cgroup separately and aggregates it at each level of the hierarchy. 1031 This may cause non-negligible overhead for some workloads when under 1032 deep level of the hierarchy, in which case this control attribute can 1033 be used to disable PSI accounting in the non-leaf cgroups. 1034 1035 irq.pressure 1036 A read-write nested-keyed file. 1037 1038 Shows pressure stall information for IRQ/SOFTIRQ. See 1039 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1040 1041Controllers 1042=========== 1043 1044.. _cgroup-v2-cpu: 1045 1046CPU 1047--- 1048 1049The "cpu" controllers regulates distribution of CPU cycles. This 1050controller implements weight and absolute bandwidth limit models for 1051normal scheduling policy and absolute bandwidth allocation model for 1052realtime scheduling policy. 1053 1054In all the above models, cycles distribution is defined only on a temporal 1055base and it does not account for the frequency at which tasks are executed. 1056The (optional) utilization clamping support allows to hint the schedutil 1057cpufreq governor about the minimum desired frequency which should always be 1058provided by a CPU, as well as the maximum desired frequency, which should not 1059be exceeded by a CPU. 1060 1061WARNING: cgroup2 doesn't yet support control of realtime processes and 1062the cpu controller can only be enabled when all RT processes are in 1063the root cgroup. Be aware that system management software may already 1064have placed RT processes into nonroot cgroups during the system boot 1065process, and these processes may need to be moved to the root cgroup 1066before the cpu controller can be enabled. 1067 1068 1069CPU Interface Files 1070~~~~~~~~~~~~~~~~~~~ 1071 1072All time durations are in microseconds. 1073 1074 cpu.stat 1075 A read-only flat-keyed file. 1076 This file exists whether the controller is enabled or not. 1077 1078 It always reports the following three stats: 1079 1080 - usage_usec 1081 - user_usec 1082 - system_usec 1083 1084 and the following five when the controller is enabled: 1085 1086 - nr_periods 1087 - nr_throttled 1088 - throttled_usec 1089 - nr_bursts 1090 - burst_usec 1091 1092 cpu.weight 1093 A read-write single value file which exists on non-root 1094 cgroups. The default is "100". 1095 1096 For non idle groups (cpu.idle = 0), the weight is in the 1097 range [1, 10000]. 1098 1099 If the cgroup has been configured to be SCHED_IDLE (cpu.idle = 1), 1100 then the weight will show as a 0. 1101 1102 cpu.weight.nice 1103 A read-write single value file which exists on non-root 1104 cgroups. The default is "0". 1105 1106 The nice value is in the range [-20, 19]. 1107 1108 This interface file is an alternative interface for 1109 "cpu.weight" and allows reading and setting weight using the 1110 same values used by nice(2). Because the range is smaller and 1111 granularity is coarser for the nice values, the read value is 1112 the closest approximation of the current weight. 1113 1114 cpu.max 1115 A read-write two value file which exists on non-root cgroups. 1116 The default is "max 100000". 1117 1118 The maximum bandwidth limit. It's in the following format:: 1119 1120 $MAX $PERIOD 1121 1122 which indicates that the group may consume up to $MAX in each 1123 $PERIOD duration. "max" for $MAX indicates no limit. If only 1124 one number is written, $MAX is updated. 1125 1126 cpu.max.burst 1127 A read-write single value file which exists on non-root 1128 cgroups. The default is "0". 1129 1130 The burst in the range [0, $MAX]. 1131 1132 cpu.pressure 1133 A read-write nested-keyed file. 1134 1135 Shows pressure stall information for CPU. See 1136 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1137 1138 cpu.uclamp.min 1139 A read-write single value file which exists on non-root cgroups. 1140 The default is "0", i.e. no utilization boosting. 1141 1142 The requested minimum utilization (protection) as a percentage 1143 rational number, e.g. 12.34 for 12.34%. 1144 1145 This interface allows reading and setting minimum utilization clamp 1146 values similar to the sched_setattr(2). This minimum utilization 1147 value is used to clamp the task specific minimum utilization clamp. 1148 1149 The requested minimum utilization (protection) is always capped by 1150 the current value for the maximum utilization (limit), i.e. 1151 `cpu.uclamp.max`. 1152 1153 cpu.uclamp.max 1154 A read-write single value file which exists on non-root cgroups. 1155 The default is "max". i.e. no utilization capping 1156 1157 The requested maximum utilization (limit) as a percentage rational 1158 number, e.g. 98.76 for 98.76%. 1159 1160 This interface allows reading and setting maximum utilization clamp 1161 values similar to the sched_setattr(2). This maximum utilization 1162 value is used to clamp the task specific maximum utilization clamp. 1163 1164 cpu.idle 1165 A read-write single value file which exists on non-root cgroups. 1166 The default is 0. 1167 1168 This is the cgroup analog of the per-task SCHED_IDLE sched policy. 1169 Setting this value to a 1 will make the scheduling policy of the 1170 cgroup SCHED_IDLE. The threads inside the cgroup will retain their 1171 own relative priorities, but the cgroup itself will be treated as 1172 very low priority relative to its peers. 1173 1174 1175 1176Memory 1177------ 1178 1179The "memory" controller regulates distribution of memory. Memory is 1180stateful and implements both limit and protection models. Due to the 1181intertwining between memory usage and reclaim pressure and the 1182stateful nature of memory, the distribution model is relatively 1183complex. 1184 1185While not completely water-tight, all major memory usages by a given 1186cgroup are tracked so that the total memory consumption can be 1187accounted and controlled to a reasonable extent. Currently, the 1188following types of memory usages are tracked. 1189 1190- Userland memory - page cache and anonymous memory. 1191 1192- Kernel data structures such as dentries and inodes. 1193 1194- TCP socket buffers. 1195 1196The above list may expand in the future for better coverage. 1197 1198 1199Memory Interface Files 1200~~~~~~~~~~~~~~~~~~~~~~ 1201 1202All memory amounts are in bytes. If a value which is not aligned to 1203PAGE_SIZE is written, the value may be rounded up to the closest 1204PAGE_SIZE multiple when read back. 1205 1206 memory.current 1207 A read-only single value file which exists on non-root 1208 cgroups. 1209 1210 The total amount of memory currently being used by the cgroup 1211 and its descendants. 1212 1213 memory.min 1214 A read-write single value file which exists on non-root 1215 cgroups. The default is "0". 1216 1217 Hard memory protection. If the memory usage of a cgroup 1218 is within its effective min boundary, the cgroup's memory 1219 won't be reclaimed under any conditions. If there is no 1220 unprotected reclaimable memory available, OOM killer 1221 is invoked. Above the effective min boundary (or 1222 effective low boundary if it is higher), pages are reclaimed 1223 proportionally to the overage, reducing reclaim pressure for 1224 smaller overages. 1225 1226 Effective min boundary is limited by memory.min values of 1227 all ancestor cgroups. If there is memory.min overcommitment 1228 (child cgroup or cgroups are requiring more protected memory 1229 than parent will allow), then each child cgroup will get 1230 the part of parent's protection proportional to its 1231 actual memory usage below memory.min. 1232 1233 Putting more memory than generally available under this 1234 protection is discouraged and may lead to constant OOMs. 1235 1236 If a memory cgroup is not populated with processes, 1237 its memory.min is ignored. 1238 1239 memory.low 1240 A read-write single value file which exists on non-root 1241 cgroups. The default is "0". 1242 1243 Best-effort memory protection. If the memory usage of a 1244 cgroup is within its effective low boundary, the cgroup's 1245 memory won't be reclaimed unless there is no reclaimable 1246 memory available in unprotected cgroups. 1247 Above the effective low boundary (or 1248 effective min boundary if it is higher), pages are reclaimed 1249 proportionally to the overage, reducing reclaim pressure for 1250 smaller overages. 1251 1252 Effective low boundary is limited by memory.low values of 1253 all ancestor cgroups. If there is memory.low overcommitment 1254 (child cgroup or cgroups are requiring more protected memory 1255 than parent will allow), then each child cgroup will get 1256 the part of parent's protection proportional to its 1257 actual memory usage below memory.low. 1258 1259 Putting more memory than generally available under this 1260 protection is discouraged. 1261 1262 memory.high 1263 A read-write single value file which exists on non-root 1264 cgroups. The default is "max". 1265 1266 Memory usage throttle limit. If a cgroup's usage goes 1267 over the high boundary, the processes of the cgroup are 1268 throttled and put under heavy reclaim pressure. 1269 1270 Going over the high limit never invokes the OOM killer and 1271 under extreme conditions the limit may be breached. The high 1272 limit should be used in scenarios where an external process 1273 monitors the limited cgroup to alleviate heavy reclaim 1274 pressure. 1275 1276 memory.max 1277 A read-write single value file which exists on non-root 1278 cgroups. The default is "max". 1279 1280 Memory usage hard limit. This is the main mechanism to limit 1281 memory usage of a cgroup. If a cgroup's memory usage reaches 1282 this limit and can't be reduced, the OOM killer is invoked in 1283 the cgroup. Under certain circumstances, the usage may go 1284 over the limit temporarily. 1285 1286 In default configuration regular 0-order allocations always 1287 succeed unless OOM killer chooses current task as a victim. 1288 1289 Some kinds of allocations don't invoke the OOM killer. 1290 Caller could retry them differently, return into userspace 1291 as -ENOMEM or silently ignore in cases like disk readahead. 1292 1293 memory.reclaim 1294 A write-only nested-keyed file which exists for all cgroups. 1295 1296 This is a simple interface to trigger memory reclaim in the 1297 target cgroup. 1298 1299 This file accepts a single key, the number of bytes to reclaim. 1300 No nested keys are currently supported. 1301 1302 Example:: 1303 1304 echo "1G" > memory.reclaim 1305 1306 The interface can be later extended with nested keys to 1307 configure the reclaim behavior. For example, specify the 1308 type of memory to reclaim from (anon, file, ..). 1309 1310 Please note that the kernel can over or under reclaim from 1311 the target cgroup. If less bytes are reclaimed than the 1312 specified amount, -EAGAIN is returned. 1313 1314 Please note that the proactive reclaim (triggered by this 1315 interface) is not meant to indicate memory pressure on the 1316 memory cgroup. Therefore socket memory balancing triggered by 1317 the memory reclaim normally is not exercised in this case. 1318 This means that the networking layer will not adapt based on 1319 reclaim induced by memory.reclaim. 1320 1321 memory.peak 1322 A read-only single value file which exists on non-root 1323 cgroups. 1324 1325 The max memory usage recorded for the cgroup and its 1326 descendants since the creation of the cgroup. 1327 1328 memory.oom.group 1329 A read-write single value file which exists on non-root 1330 cgroups. The default value is "0". 1331 1332 Determines whether the cgroup should be treated as 1333 an indivisible workload by the OOM killer. If set, 1334 all tasks belonging to the cgroup or to its descendants 1335 (if the memory cgroup is not a leaf cgroup) are killed 1336 together or not at all. This can be used to avoid 1337 partial kills to guarantee workload integrity. 1338 1339 Tasks with the OOM protection (oom_score_adj set to -1000) 1340 are treated as an exception and are never killed. 1341 1342 If the OOM killer is invoked in a cgroup, it's not going 1343 to kill any tasks outside of this cgroup, regardless 1344 memory.oom.group values of ancestor cgroups. 1345 1346 memory.events 1347 A read-only flat-keyed file which exists on non-root cgroups. 1348 The following entries are defined. Unless specified 1349 otherwise, a value change in this file generates a file 1350 modified event. 1351 1352 Note that all fields in this file are hierarchical and the 1353 file modified event can be generated due to an event down the 1354 hierarchy. For the local events at the cgroup level see 1355 memory.events.local. 1356 1357 low 1358 The number of times the cgroup is reclaimed due to 1359 high memory pressure even though its usage is under 1360 the low boundary. This usually indicates that the low 1361 boundary is over-committed. 1362 1363 high 1364 The number of times processes of the cgroup are 1365 throttled and routed to perform direct memory reclaim 1366 because the high memory boundary was exceeded. For a 1367 cgroup whose memory usage is capped by the high limit 1368 rather than global memory pressure, this event's 1369 occurrences are expected. 1370 1371 max 1372 The number of times the cgroup's memory usage was 1373 about to go over the max boundary. If direct reclaim 1374 fails to bring it down, the cgroup goes to OOM state. 1375 1376 oom 1377 The number of time the cgroup's memory usage was 1378 reached the limit and allocation was about to fail. 1379 1380 This event is not raised if the OOM killer is not 1381 considered as an option, e.g. for failed high-order 1382 allocations or if caller asked to not retry attempts. 1383 1384 oom_kill 1385 The number of processes belonging to this cgroup 1386 killed by any kind of OOM killer. 1387 1388 oom_group_kill 1389 The number of times a group OOM has occurred. 1390 1391 memory.events.local 1392 Similar to memory.events but the fields in the file are local 1393 to the cgroup i.e. not hierarchical. The file modified event 1394 generated on this file reflects only the local events. 1395 1396 memory.stat 1397 A read-only flat-keyed file which exists on non-root cgroups. 1398 1399 This breaks down the cgroup's memory footprint into different 1400 types of memory, type-specific details, and other information 1401 on the state and past events of the memory management system. 1402 1403 All memory amounts are in bytes. 1404 1405 The entries are ordered to be human readable, and new entries 1406 can show up in the middle. Don't rely on items remaining in a 1407 fixed position; use the keys to look up specific values! 1408 1409 If the entry has no per-node counter (or not show in the 1410 memory.numa_stat). We use 'npn' (non-per-node) as the tag 1411 to indicate that it will not show in the memory.numa_stat. 1412 1413 anon 1414 Amount of memory used in anonymous mappings such as 1415 brk(), sbrk(), and mmap(MAP_ANONYMOUS) 1416 1417 file 1418 Amount of memory used to cache filesystem data, 1419 including tmpfs and shared memory. 1420 1421 kernel (npn) 1422 Amount of total kernel memory, including 1423 (kernel_stack, pagetables, percpu, vmalloc, slab) in 1424 addition to other kernel memory use cases. 1425 1426 kernel_stack 1427 Amount of memory allocated to kernel stacks. 1428 1429 pagetables 1430 Amount of memory allocated for page tables. 1431 1432 sec_pagetables 1433 Amount of memory allocated for secondary page tables, 1434 this currently includes KVM mmu allocations on x86 1435 and arm64. 1436 1437 percpu (npn) 1438 Amount of memory used for storing per-cpu kernel 1439 data structures. 1440 1441 sock (npn) 1442 Amount of memory used in network transmission buffers 1443 1444 vmalloc (npn) 1445 Amount of memory used for vmap backed memory. 1446 1447 shmem 1448 Amount of cached filesystem data that is swap-backed, 1449 such as tmpfs, shm segments, shared anonymous mmap()s 1450 1451 zswap 1452 Amount of memory consumed by the zswap compression backend. 1453 1454 zswapped 1455 Amount of application memory swapped out to zswap. 1456 1457 file_mapped 1458 Amount of cached filesystem data mapped with mmap() 1459 1460 file_dirty 1461 Amount of cached filesystem data that was modified but 1462 not yet written back to disk 1463 1464 file_writeback 1465 Amount of cached filesystem data that was modified and 1466 is currently being written back to disk 1467 1468 swapcached 1469 Amount of swap cached in memory. The swapcache is accounted 1470 against both memory and swap usage. 1471 1472 anon_thp 1473 Amount of memory used in anonymous mappings backed by 1474 transparent hugepages 1475 1476 file_thp 1477 Amount of cached filesystem data backed by transparent 1478 hugepages 1479 1480 shmem_thp 1481 Amount of shm, tmpfs, shared anonymous mmap()s backed by 1482 transparent hugepages 1483 1484 inactive_anon, active_anon, inactive_file, active_file, unevictable 1485 Amount of memory, swap-backed and filesystem-backed, 1486 on the internal memory management lists used by the 1487 page reclaim algorithm. 1488 1489 As these represent internal list state (eg. shmem pages are on anon 1490 memory management lists), inactive_foo + active_foo may not be equal to 1491 the value for the foo counter, since the foo counter is type-based, not 1492 list-based. 1493 1494 slab_reclaimable 1495 Part of "slab" that might be reclaimed, such as 1496 dentries and inodes. 1497 1498 slab_unreclaimable 1499 Part of "slab" that cannot be reclaimed on memory 1500 pressure. 1501 1502 slab (npn) 1503 Amount of memory used for storing in-kernel data 1504 structures. 1505 1506 workingset_refault_anon 1507 Number of refaults of previously evicted anonymous pages. 1508 1509 workingset_refault_file 1510 Number of refaults of previously evicted file pages. 1511 1512 workingset_activate_anon 1513 Number of refaulted anonymous pages that were immediately 1514 activated. 1515 1516 workingset_activate_file 1517 Number of refaulted file pages that were immediately activated. 1518 1519 workingset_restore_anon 1520 Number of restored anonymous pages which have been detected as 1521 an active workingset before they got reclaimed. 1522 1523 workingset_restore_file 1524 Number of restored file pages which have been detected as an 1525 active workingset before they got reclaimed. 1526 1527 workingset_nodereclaim 1528 Number of times a shadow node has been reclaimed 1529 1530 pgscan (npn) 1531 Amount of scanned pages (in an inactive LRU list) 1532 1533 pgsteal (npn) 1534 Amount of reclaimed pages 1535 1536 pgscan_kswapd (npn) 1537 Amount of scanned pages by kswapd (in an inactive LRU list) 1538 1539 pgscan_direct (npn) 1540 Amount of scanned pages directly (in an inactive LRU list) 1541 1542 pgscan_khugepaged (npn) 1543 Amount of scanned pages by khugepaged (in an inactive LRU list) 1544 1545 pgsteal_kswapd (npn) 1546 Amount of reclaimed pages by kswapd 1547 1548 pgsteal_direct (npn) 1549 Amount of reclaimed pages directly 1550 1551 pgsteal_khugepaged (npn) 1552 Amount of reclaimed pages by khugepaged 1553 1554 pgfault (npn) 1555 Total number of page faults incurred 1556 1557 pgmajfault (npn) 1558 Number of major page faults incurred 1559 1560 pgrefill (npn) 1561 Amount of scanned pages (in an active LRU list) 1562 1563 pgactivate (npn) 1564 Amount of pages moved to the active LRU list 1565 1566 pgdeactivate (npn) 1567 Amount of pages moved to the inactive LRU list 1568 1569 pglazyfree (npn) 1570 Amount of pages postponed to be freed under memory pressure 1571 1572 pglazyfreed (npn) 1573 Amount of reclaimed lazyfree pages 1574 1575 zswpin 1576 Number of pages moved in to memory from zswap. 1577 1578 zswpout 1579 Number of pages moved out of memory to zswap. 1580 1581 zswpwb 1582 Number of pages written from zswap to swap. 1583 1584 thp_fault_alloc (npn) 1585 Number of transparent hugepages which were allocated to satisfy 1586 a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE 1587 is not set. 1588 1589 thp_collapse_alloc (npn) 1590 Number of transparent hugepages which were allocated to allow 1591 collapsing an existing range of pages. This counter is not 1592 present when CONFIG_TRANSPARENT_HUGEPAGE is not set. 1593 1594 thp_swpout (npn) 1595 Number of transparent hugepages which are swapout in one piece 1596 without splitting. 1597 1598 thp_swpout_fallback (npn) 1599 Number of transparent hugepages which were split before swapout. 1600 Usually because failed to allocate some continuous swap space 1601 for the huge page. 1602 1603 memory.numa_stat 1604 A read-only nested-keyed file which exists on non-root cgroups. 1605 1606 This breaks down the cgroup's memory footprint into different 1607 types of memory, type-specific details, and other information 1608 per node on the state of the memory management system. 1609 1610 This is useful for providing visibility into the NUMA locality 1611 information within an memcg since the pages are allowed to be 1612 allocated from any physical node. One of the use case is evaluating 1613 application performance by combining this information with the 1614 application's CPU allocation. 1615 1616 All memory amounts are in bytes. 1617 1618 The output format of memory.numa_stat is:: 1619 1620 type N0=<bytes in node 0> N1=<bytes in node 1> ... 1621 1622 The entries are ordered to be human readable, and new entries 1623 can show up in the middle. Don't rely on items remaining in a 1624 fixed position; use the keys to look up specific values! 1625 1626 The entries can refer to the memory.stat. 1627 1628 memory.swap.current 1629 A read-only single value file which exists on non-root 1630 cgroups. 1631 1632 The total amount of swap currently being used by the cgroup 1633 and its descendants. 1634 1635 memory.swap.high 1636 A read-write single value file which exists on non-root 1637 cgroups. The default is "max". 1638 1639 Swap usage throttle limit. If a cgroup's swap usage exceeds 1640 this limit, all its further allocations will be throttled to 1641 allow userspace to implement custom out-of-memory procedures. 1642 1643 This limit marks a point of no return for the cgroup. It is NOT 1644 designed to manage the amount of swapping a workload does 1645 during regular operation. Compare to memory.swap.max, which 1646 prohibits swapping past a set amount, but lets the cgroup 1647 continue unimpeded as long as other memory can be reclaimed. 1648 1649 Healthy workloads are not expected to reach this limit. 1650 1651 memory.swap.peak 1652 A read-only single value file which exists on non-root 1653 cgroups. 1654 1655 The max swap usage recorded for the cgroup and its 1656 descendants since the creation of the cgroup. 1657 1658 memory.swap.max 1659 A read-write single value file which exists on non-root 1660 cgroups. The default is "max". 1661 1662 Swap usage hard limit. If a cgroup's swap usage reaches this 1663 limit, anonymous memory of the cgroup will not be swapped out. 1664 1665 memory.swap.events 1666 A read-only flat-keyed file which exists on non-root cgroups. 1667 The following entries are defined. Unless specified 1668 otherwise, a value change in this file generates a file 1669 modified event. 1670 1671 high 1672 The number of times the cgroup's swap usage was over 1673 the high threshold. 1674 1675 max 1676 The number of times the cgroup's swap usage was about 1677 to go over the max boundary and swap allocation 1678 failed. 1679 1680 fail 1681 The number of times swap allocation failed either 1682 because of running out of swap system-wide or max 1683 limit. 1684 1685 When reduced under the current usage, the existing swap 1686 entries are reclaimed gradually and the swap usage may stay 1687 higher than the limit for an extended period of time. This 1688 reduces the impact on the workload and memory management. 1689 1690 memory.zswap.current 1691 A read-only single value file which exists on non-root 1692 cgroups. 1693 1694 The total amount of memory consumed by the zswap compression 1695 backend. 1696 1697 memory.zswap.max 1698 A read-write single value file which exists on non-root 1699 cgroups. The default is "max". 1700 1701 Zswap usage hard limit. If a cgroup's zswap pool reaches this 1702 limit, it will refuse to take any more stores before existing 1703 entries fault back in or are written out to disk. 1704 1705 memory.zswap.writeback 1706 A read-write single value file. The default value is "1". The 1707 initial value of the root cgroup is 1, and when a new cgroup is 1708 created, it inherits the current value of its parent. 1709 1710 When this is set to 0, all swapping attempts to swapping devices 1711 are disabled. This included both zswap writebacks, and swapping due 1712 to zswap store failures. If the zswap store failures are recurring 1713 (for e.g if the pages are incompressible), users can observe 1714 reclaim inefficiency after disabling writeback (because the same 1715 pages might be rejected again and again). 1716 1717 Note that this is subtly different from setting memory.swap.max to 1718 0, as it still allows for pages to be written to the zswap pool. 1719 1720 memory.pressure 1721 A read-only nested-keyed file. 1722 1723 Shows pressure stall information for memory. See 1724 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1725 1726 1727Usage Guidelines 1728~~~~~~~~~~~~~~~~ 1729 1730"memory.high" is the main mechanism to control memory usage. 1731Over-committing on high limit (sum of high limits > available memory) 1732and letting global memory pressure to distribute memory according to 1733usage is a viable strategy. 1734 1735Because breach of the high limit doesn't trigger the OOM killer but 1736throttles the offending cgroup, a management agent has ample 1737opportunities to monitor and take appropriate actions such as granting 1738more memory or terminating the workload. 1739 1740Determining whether a cgroup has enough memory is not trivial as 1741memory usage doesn't indicate whether the workload can benefit from 1742more memory. For example, a workload which writes data received from 1743network to a file can use all available memory but can also operate as 1744performant with a small amount of memory. A measure of memory 1745pressure - how much the workload is being impacted due to lack of 1746memory - is necessary to determine whether a workload needs more 1747memory; unfortunately, memory pressure monitoring mechanism isn't 1748implemented yet. 1749 1750 1751Memory Ownership 1752~~~~~~~~~~~~~~~~ 1753 1754A memory area is charged to the cgroup which instantiated it and stays 1755charged to the cgroup until the area is released. Migrating a process 1756to a different cgroup doesn't move the memory usages that it 1757instantiated while in the previous cgroup to the new cgroup. 1758 1759A memory area may be used by processes belonging to different cgroups. 1760To which cgroup the area will be charged is in-deterministic; however, 1761over time, the memory area is likely to end up in a cgroup which has 1762enough memory allowance to avoid high reclaim pressure. 1763 1764If a cgroup sweeps a considerable amount of memory which is expected 1765to be accessed repeatedly by other cgroups, it may make sense to use 1766POSIX_FADV_DONTNEED to relinquish the ownership of memory areas 1767belonging to the affected files to ensure correct memory ownership. 1768 1769 1770IO 1771-- 1772 1773The "io" controller regulates the distribution of IO resources. This 1774controller implements both weight based and absolute bandwidth or IOPS 1775limit distribution; however, weight based distribution is available 1776only if cfq-iosched is in use and neither scheme is available for 1777blk-mq devices. 1778 1779 1780IO Interface Files 1781~~~~~~~~~~~~~~~~~~ 1782 1783 io.stat 1784 A read-only nested-keyed file. 1785 1786 Lines are keyed by $MAJ:$MIN device numbers and not ordered. 1787 The following nested keys are defined. 1788 1789 ====== ===================== 1790 rbytes Bytes read 1791 wbytes Bytes written 1792 rios Number of read IOs 1793 wios Number of write IOs 1794 dbytes Bytes discarded 1795 dios Number of discard IOs 1796 ====== ===================== 1797 1798 An example read output follows:: 1799 1800 8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0 1801 8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021 1802 1803 io.cost.qos 1804 A read-write nested-keyed file which exists only on the root 1805 cgroup. 1806 1807 This file configures the Quality of Service of the IO cost 1808 model based controller (CONFIG_BLK_CGROUP_IOCOST) which 1809 currently implements "io.weight" proportional control. Lines 1810 are keyed by $MAJ:$MIN device numbers and not ordered. The 1811 line for a given device is populated on the first write for 1812 the device on "io.cost.qos" or "io.cost.model". The following 1813 nested keys are defined. 1814 1815 ====== ===================================== 1816 enable Weight-based control enable 1817 ctrl "auto" or "user" 1818 rpct Read latency percentile [0, 100] 1819 rlat Read latency threshold 1820 wpct Write latency percentile [0, 100] 1821 wlat Write latency threshold 1822 min Minimum scaling percentage [1, 10000] 1823 max Maximum scaling percentage [1, 10000] 1824 ====== ===================================== 1825 1826 The controller is disabled by default and can be enabled by 1827 setting "enable" to 1. "rpct" and "wpct" parameters default 1828 to zero and the controller uses internal device saturation 1829 state to adjust the overall IO rate between "min" and "max". 1830 1831 When a better control quality is needed, latency QoS 1832 parameters can be configured. For example:: 1833 1834 8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0 1835 1836 shows that on sdb, the controller is enabled, will consider 1837 the device saturated if the 95th percentile of read completion 1838 latencies is above 75ms or write 150ms, and adjust the overall 1839 IO issue rate between 50% and 150% accordingly. 1840 1841 The lower the saturation point, the better the latency QoS at 1842 the cost of aggregate bandwidth. The narrower the allowed 1843 adjustment range between "min" and "max", the more conformant 1844 to the cost model the IO behavior. Note that the IO issue 1845 base rate may be far off from 100% and setting "min" and "max" 1846 blindly can lead to a significant loss of device capacity or 1847 control quality. "min" and "max" are useful for regulating 1848 devices which show wide temporary behavior changes - e.g. a 1849 ssd which accepts writes at the line speed for a while and 1850 then completely stalls for multiple seconds. 1851 1852 When "ctrl" is "auto", the parameters are controlled by the 1853 kernel and may change automatically. Setting "ctrl" to "user" 1854 or setting any of the percentile and latency parameters puts 1855 it into "user" mode and disables the automatic changes. The 1856 automatic mode can be restored by setting "ctrl" to "auto". 1857 1858 io.cost.model 1859 A read-write nested-keyed file which exists only on the root 1860 cgroup. 1861 1862 This file configures the cost model of the IO cost model based 1863 controller (CONFIG_BLK_CGROUP_IOCOST) which currently 1864 implements "io.weight" proportional control. Lines are keyed 1865 by $MAJ:$MIN device numbers and not ordered. The line for a 1866 given device is populated on the first write for the device on 1867 "io.cost.qos" or "io.cost.model". The following nested keys 1868 are defined. 1869 1870 ===== ================================ 1871 ctrl "auto" or "user" 1872 model The cost model in use - "linear" 1873 ===== ================================ 1874 1875 When "ctrl" is "auto", the kernel may change all parameters 1876 dynamically. When "ctrl" is set to "user" or any other 1877 parameters are written to, "ctrl" become "user" and the 1878 automatic changes are disabled. 1879 1880 When "model" is "linear", the following model parameters are 1881 defined. 1882 1883 ============= ======================================== 1884 [r|w]bps The maximum sequential IO throughput 1885 [r|w]seqiops The maximum 4k sequential IOs per second 1886 [r|w]randiops The maximum 4k random IOs per second 1887 ============= ======================================== 1888 1889 From the above, the builtin linear model determines the base 1890 costs of a sequential and random IO and the cost coefficient 1891 for the IO size. While simple, this model can cover most 1892 common device classes acceptably. 1893 1894 The IO cost model isn't expected to be accurate in absolute 1895 sense and is scaled to the device behavior dynamically. 1896 1897 If needed, tools/cgroup/iocost_coef_gen.py can be used to 1898 generate device-specific coefficients. 1899 1900 io.weight 1901 A read-write flat-keyed file which exists on non-root cgroups. 1902 The default is "default 100". 1903 1904 The first line is the default weight applied to devices 1905 without specific override. The rest are overrides keyed by 1906 $MAJ:$MIN device numbers and not ordered. The weights are in 1907 the range [1, 10000] and specifies the relative amount IO time 1908 the cgroup can use in relation to its siblings. 1909 1910 The default weight can be updated by writing either "default 1911 $WEIGHT" or simply "$WEIGHT". Overrides can be set by writing 1912 "$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default". 1913 1914 An example read output follows:: 1915 1916 default 100 1917 8:16 200 1918 8:0 50 1919 1920 io.max 1921 A read-write nested-keyed file which exists on non-root 1922 cgroups. 1923 1924 BPS and IOPS based IO limit. Lines are keyed by $MAJ:$MIN 1925 device numbers and not ordered. The following nested keys are 1926 defined. 1927 1928 ===== ================================== 1929 rbps Max read bytes per second 1930 wbps Max write bytes per second 1931 riops Max read IO operations per second 1932 wiops Max write IO operations per second 1933 ===== ================================== 1934 1935 When writing, any number of nested key-value pairs can be 1936 specified in any order. "max" can be specified as the value 1937 to remove a specific limit. If the same key is specified 1938 multiple times, the outcome is undefined. 1939 1940 BPS and IOPS are measured in each IO direction and IOs are 1941 delayed if limit is reached. Temporary bursts are allowed. 1942 1943 Setting read limit at 2M BPS and write at 120 IOPS for 8:16:: 1944 1945 echo "8:16 rbps=2097152 wiops=120" > io.max 1946 1947 Reading returns the following:: 1948 1949 8:16 rbps=2097152 wbps=max riops=max wiops=120 1950 1951 Write IOPS limit can be removed by writing the following:: 1952 1953 echo "8:16 wiops=max" > io.max 1954 1955 Reading now returns the following:: 1956 1957 8:16 rbps=2097152 wbps=max riops=max wiops=max 1958 1959 io.pressure 1960 A read-only nested-keyed file. 1961 1962 Shows pressure stall information for IO. See 1963 :ref:`Documentation/accounting/psi.rst <psi>` for details. 1964 1965 1966Writeback 1967~~~~~~~~~ 1968 1969Page cache is dirtied through buffered writes and shared mmaps and 1970written asynchronously to the backing filesystem by the writeback 1971mechanism. Writeback sits between the memory and IO domains and 1972regulates the proportion of dirty memory by balancing dirtying and 1973write IOs. 1974 1975The io controller, in conjunction with the memory controller, 1976implements control of page cache writeback IOs. The memory controller 1977defines the memory domain that dirty memory ratio is calculated and 1978maintained for and the io controller defines the io domain which 1979writes out dirty pages for the memory domain. Both system-wide and 1980per-cgroup dirty memory states are examined and the more restrictive 1981of the two is enforced. 1982 1983cgroup writeback requires explicit support from the underlying 1984filesystem. Currently, cgroup writeback is implemented on ext2, ext4, 1985btrfs, f2fs, and xfs. On other filesystems, all writeback IOs are 1986attributed to the root cgroup. 1987 1988There are inherent differences in memory and writeback management 1989which affects how cgroup ownership is tracked. Memory is tracked per 1990page while writeback per inode. For the purpose of writeback, an 1991inode is assigned to a cgroup and all IO requests to write dirty pages 1992from the inode are attributed to that cgroup. 1993 1994As cgroup ownership for memory is tracked per page, there can be pages 1995which are associated with different cgroups than the one the inode is 1996associated with. These are called foreign pages. The writeback 1997constantly keeps track of foreign pages and, if a particular foreign 1998cgroup becomes the majority over a certain period of time, switches 1999the ownership of the inode to that cgroup. 2000 2001While this model is enough for most use cases where a given inode is 2002mostly dirtied by a single cgroup even when the main writing cgroup 2003changes over time, use cases where multiple cgroups write to a single 2004inode simultaneously are not supported well. In such circumstances, a 2005significant portion of IOs are likely to be attributed incorrectly. 2006As memory controller assigns page ownership on the first use and 2007doesn't update it until the page is released, even if writeback 2008strictly follows page ownership, multiple cgroups dirtying overlapping 2009areas wouldn't work as expected. It's recommended to avoid such usage 2010patterns. 2011 2012The sysctl knobs which affect writeback behavior are applied to cgroup 2013writeback as follows. 2014 2015 vm.dirty_background_ratio, vm.dirty_ratio 2016 These ratios apply the same to cgroup writeback with the 2017 amount of available memory capped by limits imposed by the 2018 memory controller and system-wide clean memory. 2019 2020 vm.dirty_background_bytes, vm.dirty_bytes 2021 For cgroup writeback, this is calculated into ratio against 2022 total available memory and applied the same way as 2023 vm.dirty[_background]_ratio. 2024 2025 2026IO Latency 2027~~~~~~~~~~ 2028 2029This is a cgroup v2 controller for IO workload protection. You provide a group 2030with a latency target, and if the average latency exceeds that target the 2031controller will throttle any peers that have a lower latency target than the 2032protected workload. 2033 2034The limits are only applied at the peer level in the hierarchy. This means that 2035in the diagram below, only groups A, B, and C will influence each other, and 2036groups D and F will influence each other. Group G will influence nobody:: 2037 2038 [root] 2039 / | \ 2040 A B C 2041 / \ | 2042 D F G 2043 2044 2045So the ideal way to configure this is to set io.latency in groups A, B, and C. 2046Generally you do not want to set a value lower than the latency your device 2047supports. Experiment to find the value that works best for your workload. 2048Start at higher than the expected latency for your device and watch the 2049avg_lat value in io.stat for your workload group to get an idea of the 2050latency you see during normal operation. Use the avg_lat value as a basis for 2051your real setting, setting at 10-15% higher than the value in io.stat. 2052 2053How IO Latency Throttling Works 2054~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2055 2056io.latency is work conserving; so as long as everybody is meeting their latency 2057target the controller doesn't do anything. Once a group starts missing its 2058target it begins throttling any peer group that has a higher target than itself. 2059This throttling takes 2 forms: 2060 2061- Queue depth throttling. This is the number of outstanding IO's a group is 2062 allowed to have. We will clamp down relatively quickly, starting at no limit 2063 and going all the way down to 1 IO at a time. 2064 2065- Artificial delay induction. There are certain types of IO that cannot be 2066 throttled without possibly adversely affecting higher priority groups. This 2067 includes swapping and metadata IO. These types of IO are allowed to occur 2068 normally, however they are "charged" to the originating group. If the 2069 originating group is being throttled you will see the use_delay and delay 2070 fields in io.stat increase. The delay value is how many microseconds that are 2071 being added to any process that runs in this group. Because this number can 2072 grow quite large if there is a lot of swapping or metadata IO occurring we 2073 limit the individual delay events to 1 second at a time. 2074 2075Once the victimized group starts meeting its latency target again it will start 2076unthrottling any peer groups that were throttled previously. If the victimized 2077group simply stops doing IO the global counter will unthrottle appropriately. 2078 2079IO Latency Interface Files 2080~~~~~~~~~~~~~~~~~~~~~~~~~~ 2081 2082 io.latency 2083 This takes a similar format as the other controllers. 2084 2085 "MAJOR:MINOR target=<target time in microseconds>" 2086 2087 io.stat 2088 If the controller is enabled you will see extra stats in io.stat in 2089 addition to the normal ones. 2090 2091 depth 2092 This is the current queue depth for the group. 2093 2094 avg_lat 2095 This is an exponential moving average with a decay rate of 1/exp 2096 bound by the sampling interval. The decay rate interval can be 2097 calculated by multiplying the win value in io.stat by the 2098 corresponding number of samples based on the win value. 2099 2100 win 2101 The sampling window size in milliseconds. This is the minimum 2102 duration of time between evaluation events. Windows only elapse 2103 with IO activity. Idle periods extend the most recent window. 2104 2105IO Priority 2106~~~~~~~~~~~ 2107 2108A single attribute controls the behavior of the I/O priority cgroup policy, 2109namely the io.prio.class attribute. The following values are accepted for 2110that attribute: 2111 2112 no-change 2113 Do not modify the I/O priority class. 2114 2115 promote-to-rt 2116 For requests that have a non-RT I/O priority class, change it into RT. 2117 Also change the priority level of these requests to 4. Do not modify 2118 the I/O priority of requests that have priority class RT. 2119 2120 restrict-to-be 2121 For requests that do not have an I/O priority class or that have I/O 2122 priority class RT, change it into BE. Also change the priority level 2123 of these requests to 0. Do not modify the I/O priority class of 2124 requests that have priority class IDLE. 2125 2126 idle 2127 Change the I/O priority class of all requests into IDLE, the lowest 2128 I/O priority class. 2129 2130 none-to-rt 2131 Deprecated. Just an alias for promote-to-rt. 2132 2133The following numerical values are associated with the I/O priority policies: 2134 2135+----------------+---+ 2136| no-change | 0 | 2137+----------------+---+ 2138| promote-to-rt | 1 | 2139+----------------+---+ 2140| restrict-to-be | 2 | 2141+----------------+---+ 2142| idle | 3 | 2143+----------------+---+ 2144 2145The numerical value that corresponds to each I/O priority class is as follows: 2146 2147+-------------------------------+---+ 2148| IOPRIO_CLASS_NONE | 0 | 2149+-------------------------------+---+ 2150| IOPRIO_CLASS_RT (real-time) | 1 | 2151+-------------------------------+---+ 2152| IOPRIO_CLASS_BE (best effort) | 2 | 2153+-------------------------------+---+ 2154| IOPRIO_CLASS_IDLE | 3 | 2155+-------------------------------+---+ 2156 2157The algorithm to set the I/O priority class for a request is as follows: 2158 2159- If I/O priority class policy is promote-to-rt, change the request I/O 2160 priority class to IOPRIO_CLASS_RT and change the request I/O priority 2161 level to 4. 2162- If I/O priority class policy is not promote-to-rt, translate the I/O priority 2163 class policy into a number, then change the request I/O priority class 2164 into the maximum of the I/O priority class policy number and the numerical 2165 I/O priority class. 2166 2167PID 2168--- 2169 2170The process number controller is used to allow a cgroup to stop any 2171new tasks from being fork()'d or clone()'d after a specified limit is 2172reached. 2173 2174The number of tasks in a cgroup can be exhausted in ways which other 2175controllers cannot prevent, thus warranting its own controller. For 2176example, a fork bomb is likely to exhaust the number of tasks before 2177hitting memory restrictions. 2178 2179Note that PIDs used in this controller refer to TIDs, process IDs as 2180used by the kernel. 2181 2182 2183PID Interface Files 2184~~~~~~~~~~~~~~~~~~~ 2185 2186 pids.max 2187 A read-write single value file which exists on non-root 2188 cgroups. The default is "max". 2189 2190 Hard limit of number of processes. 2191 2192 pids.current 2193 A read-only single value file which exists on all cgroups. 2194 2195 The number of processes currently in the cgroup and its 2196 descendants. 2197 2198Organisational operations are not blocked by cgroup policies, so it is 2199possible to have pids.current > pids.max. This can be done by either 2200setting the limit to be smaller than pids.current, or attaching enough 2201processes to the cgroup such that pids.current is larger than 2202pids.max. However, it is not possible to violate a cgroup PID policy 2203through fork() or clone(). These will return -EAGAIN if the creation 2204of a new process would cause a cgroup policy to be violated. 2205 2206 2207Cpuset 2208------ 2209 2210The "cpuset" controller provides a mechanism for constraining 2211the CPU and memory node placement of tasks to only the resources 2212specified in the cpuset interface files in a task's current cgroup. 2213This is especially valuable on large NUMA systems where placing jobs 2214on properly sized subsets of the systems with careful processor and 2215memory placement to reduce cross-node memory access and contention 2216can improve overall system performance. 2217 2218The "cpuset" controller is hierarchical. That means the controller 2219cannot use CPUs or memory nodes not allowed in its parent. 2220 2221 2222Cpuset Interface Files 2223~~~~~~~~~~~~~~~~~~~~~~ 2224 2225 cpuset.cpus 2226 A read-write multiple values file which exists on non-root 2227 cpuset-enabled cgroups. 2228 2229 It lists the requested CPUs to be used by tasks within this 2230 cgroup. The actual list of CPUs to be granted, however, is 2231 subjected to constraints imposed by its parent and can differ 2232 from the requested CPUs. 2233 2234 The CPU numbers are comma-separated numbers or ranges. 2235 For example:: 2236 2237 # cat cpuset.cpus 2238 0-4,6,8-10 2239 2240 An empty value indicates that the cgroup is using the same 2241 setting as the nearest cgroup ancestor with a non-empty 2242 "cpuset.cpus" or all the available CPUs if none is found. 2243 2244 The value of "cpuset.cpus" stays constant until the next update 2245 and won't be affected by any CPU hotplug events. 2246 2247 cpuset.cpus.effective 2248 A read-only multiple values file which exists on all 2249 cpuset-enabled cgroups. 2250 2251 It lists the onlined CPUs that are actually granted to this 2252 cgroup by its parent. These CPUs are allowed to be used by 2253 tasks within the current cgroup. 2254 2255 If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows 2256 all the CPUs from the parent cgroup that can be available to 2257 be used by this cgroup. Otherwise, it should be a subset of 2258 "cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus" 2259 can be granted. In this case, it will be treated just like an 2260 empty "cpuset.cpus". 2261 2262 Its value will be affected by CPU hotplug events. 2263 2264 cpuset.mems 2265 A read-write multiple values file which exists on non-root 2266 cpuset-enabled cgroups. 2267 2268 It lists the requested memory nodes to be used by tasks within 2269 this cgroup. The actual list of memory nodes granted, however, 2270 is subjected to constraints imposed by its parent and can differ 2271 from the requested memory nodes. 2272 2273 The memory node numbers are comma-separated numbers or ranges. 2274 For example:: 2275 2276 # cat cpuset.mems 2277 0-1,3 2278 2279 An empty value indicates that the cgroup is using the same 2280 setting as the nearest cgroup ancestor with a non-empty 2281 "cpuset.mems" or all the available memory nodes if none 2282 is found. 2283 2284 The value of "cpuset.mems" stays constant until the next update 2285 and won't be affected by any memory nodes hotplug events. 2286 2287 Setting a non-empty value to "cpuset.mems" causes memory of 2288 tasks within the cgroup to be migrated to the designated nodes if 2289 they are currently using memory outside of the designated nodes. 2290 2291 There is a cost for this memory migration. The migration 2292 may not be complete and some memory pages may be left behind. 2293 So it is recommended that "cpuset.mems" should be set properly 2294 before spawning new tasks into the cpuset. Even if there is 2295 a need to change "cpuset.mems" with active tasks, it shouldn't 2296 be done frequently. 2297 2298 cpuset.mems.effective 2299 A read-only multiple values file which exists on all 2300 cpuset-enabled cgroups. 2301 2302 It lists the onlined memory nodes that are actually granted to 2303 this cgroup by its parent. These memory nodes are allowed to 2304 be used by tasks within the current cgroup. 2305 2306 If "cpuset.mems" is empty, it shows all the memory nodes from the 2307 parent cgroup that will be available to be used by this cgroup. 2308 Otherwise, it should be a subset of "cpuset.mems" unless none of 2309 the memory nodes listed in "cpuset.mems" can be granted. In this 2310 case, it will be treated just like an empty "cpuset.mems". 2311 2312 Its value will be affected by memory nodes hotplug events. 2313 2314 cpuset.cpus.exclusive 2315 A read-write multiple values file which exists on non-root 2316 cpuset-enabled cgroups. 2317 2318 It lists all the exclusive CPUs that are allowed to be used 2319 to create a new cpuset partition. Its value is not used 2320 unless the cgroup becomes a valid partition root. See the 2321 "cpuset.cpus.partition" section below for a description of what 2322 a cpuset partition is. 2323 2324 When the cgroup becomes a partition root, the actual exclusive 2325 CPUs that are allocated to that partition are listed in 2326 "cpuset.cpus.exclusive.effective" which may be different 2327 from "cpuset.cpus.exclusive". If "cpuset.cpus.exclusive" 2328 has previously been set, "cpuset.cpus.exclusive.effective" 2329 is always a subset of it. 2330 2331 Users can manually set it to a value that is different from 2332 "cpuset.cpus". The only constraint in setting it is that the 2333 list of CPUs must be exclusive with respect to its sibling. 2334 2335 For a parent cgroup, any one of its exclusive CPUs can only 2336 be distributed to at most one of its child cgroups. Having an 2337 exclusive CPU appearing in two or more of its child cgroups is 2338 not allowed (the exclusivity rule). A value that violates the 2339 exclusivity rule will be rejected with a write error. 2340 2341 The root cgroup is a partition root and all its available CPUs 2342 are in its exclusive CPU set. 2343 2344 cpuset.cpus.exclusive.effective 2345 A read-only multiple values file which exists on all non-root 2346 cpuset-enabled cgroups. 2347 2348 This file shows the effective set of exclusive CPUs that 2349 can be used to create a partition root. The content of this 2350 file will always be a subset of "cpuset.cpus" and its parent's 2351 "cpuset.cpus.exclusive.effective" if its parent is not the root 2352 cgroup. It will also be a subset of "cpuset.cpus.exclusive" 2353 if it is set. If "cpuset.cpus.exclusive" is not set, it is 2354 treated to have an implicit value of "cpuset.cpus" in the 2355 formation of local partition. 2356 2357 cpuset.cpus.isolated 2358 A read-only and root cgroup only multiple values file. 2359 2360 This file shows the set of all isolated CPUs used in existing 2361 isolated partitions. It will be empty if no isolated partition 2362 is created. 2363 2364 cpuset.cpus.partition 2365 A read-write single value file which exists on non-root 2366 cpuset-enabled cgroups. This flag is owned by the parent cgroup 2367 and is not delegatable. 2368 2369 It accepts only the following input values when written to. 2370 2371 ========== ===================================== 2372 "member" Non-root member of a partition 2373 "root" Partition root 2374 "isolated" Partition root without load balancing 2375 ========== ===================================== 2376 2377 A cpuset partition is a collection of cpuset-enabled cgroups with 2378 a partition root at the top of the hierarchy and its descendants 2379 except those that are separate partition roots themselves and 2380 their descendants. A partition has exclusive access to the 2381 set of exclusive CPUs allocated to it. Other cgroups outside 2382 of that partition cannot use any CPUs in that set. 2383 2384 There are two types of partitions - local and remote. A local 2385 partition is one whose parent cgroup is also a valid partition 2386 root. A remote partition is one whose parent cgroup is not a 2387 valid partition root itself. Writing to "cpuset.cpus.exclusive" 2388 is optional for the creation of a local partition as its 2389 "cpuset.cpus.exclusive" file will assume an implicit value that 2390 is the same as "cpuset.cpus" if it is not set. Writing the 2391 proper "cpuset.cpus.exclusive" values down the cgroup hierarchy 2392 before the target partition root is mandatory for the creation 2393 of a remote partition. 2394 2395 Currently, a remote partition cannot be created under a local 2396 partition. All the ancestors of a remote partition root except 2397 the root cgroup cannot be a partition root. 2398 2399 The root cgroup is always a partition root and its state cannot 2400 be changed. All other non-root cgroups start out as "member". 2401 2402 When set to "root", the current cgroup is the root of a new 2403 partition or scheduling domain. The set of exclusive CPUs is 2404 determined by the value of its "cpuset.cpus.exclusive.effective". 2405 2406 When set to "isolated", the CPUs in that partition will be in 2407 an isolated state without any load balancing from the scheduler 2408 and excluded from the unbound workqueues. Tasks placed in such 2409 a partition with multiple CPUs should be carefully distributed 2410 and bound to each of the individual CPUs for optimal performance. 2411 2412 A partition root ("root" or "isolated") can be in one of the 2413 two possible states - valid or invalid. An invalid partition 2414 root is in a degraded state where some state information may 2415 be retained, but behaves more like a "member". 2416 2417 All possible state transitions among "member", "root" and 2418 "isolated" are allowed. 2419 2420 On read, the "cpuset.cpus.partition" file can show the following 2421 values. 2422 2423 ============================= ===================================== 2424 "member" Non-root member of a partition 2425 "root" Partition root 2426 "isolated" Partition root without load balancing 2427 "root invalid (<reason>)" Invalid partition root 2428 "isolated invalid (<reason>)" Invalid isolated partition root 2429 ============================= ===================================== 2430 2431 In the case of an invalid partition root, a descriptive string on 2432 why the partition is invalid is included within parentheses. 2433 2434 For a local partition root to be valid, the following conditions 2435 must be met. 2436 2437 1) The parent cgroup is a valid partition root. 2438 2) The "cpuset.cpus.exclusive.effective" file cannot be empty, 2439 though it may contain offline CPUs. 2440 3) The "cpuset.cpus.effective" cannot be empty unless there is 2441 no task associated with this partition. 2442 2443 For a remote partition root to be valid, all the above conditions 2444 except the first one must be met. 2445 2446 External events like hotplug or changes to "cpuset.cpus" or 2447 "cpuset.cpus.exclusive" can cause a valid partition root to 2448 become invalid and vice versa. Note that a task cannot be 2449 moved to a cgroup with empty "cpuset.cpus.effective". 2450 2451 A valid non-root parent partition may distribute out all its CPUs 2452 to its child local partitions when there is no task associated 2453 with it. 2454 2455 Care must be taken to change a valid partition root to "member" 2456 as all its child local partitions, if present, will become 2457 invalid causing disruption to tasks running in those child 2458 partitions. These inactivated partitions could be recovered if 2459 their parent is switched back to a partition root with a proper 2460 value in "cpuset.cpus" or "cpuset.cpus.exclusive". 2461 2462 Poll and inotify events are triggered whenever the state of 2463 "cpuset.cpus.partition" changes. That includes changes caused 2464 by write to "cpuset.cpus.partition", cpu hotplug or other 2465 changes that modify the validity status of the partition. 2466 This will allow user space agents to monitor unexpected changes 2467 to "cpuset.cpus.partition" without the need to do continuous 2468 polling. 2469 2470 A user can pre-configure certain CPUs to an isolated state 2471 with load balancing disabled at boot time with the "isolcpus" 2472 kernel boot command line option. If those CPUs are to be put 2473 into a partition, they have to be used in an isolated partition. 2474 2475 2476Device controller 2477----------------- 2478 2479Device controller manages access to device files. It includes both 2480creation of new device files (using mknod), and access to the 2481existing device files. 2482 2483Cgroup v2 device controller has no interface files and is implemented 2484on top of cgroup BPF. To control access to device files, a user may 2485create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach 2486them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a 2487device file, corresponding BPF programs will be executed, and depending 2488on the return value the attempt will succeed or fail with -EPERM. 2489 2490A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the 2491bpf_cgroup_dev_ctx structure, which describes the device access attempt: 2492access type (mknod/read/write) and device (type, major and minor numbers). 2493If the program returns 0, the attempt fails with -EPERM, otherwise it 2494succeeds. 2495 2496An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in 2497tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree. 2498 2499 2500RDMA 2501---- 2502 2503The "rdma" controller regulates the distribution and accounting of 2504RDMA resources. 2505 2506RDMA Interface Files 2507~~~~~~~~~~~~~~~~~~~~ 2508 2509 rdma.max 2510 A readwrite nested-keyed file that exists for all the cgroups 2511 except root that describes current configured resource limit 2512 for a RDMA/IB device. 2513 2514 Lines are keyed by device name and are not ordered. 2515 Each line contains space separated resource name and its configured 2516 limit that can be distributed. 2517 2518 The following nested keys are defined. 2519 2520 ========== ============================= 2521 hca_handle Maximum number of HCA Handles 2522 hca_object Maximum number of HCA Objects 2523 ========== ============================= 2524 2525 An example for mlx4 and ocrdma device follows:: 2526 2527 mlx4_0 hca_handle=2 hca_object=2000 2528 ocrdma1 hca_handle=3 hca_object=max 2529 2530 rdma.current 2531 A read-only file that describes current resource usage. 2532 It exists for all the cgroup except root. 2533 2534 An example for mlx4 and ocrdma device follows:: 2535 2536 mlx4_0 hca_handle=1 hca_object=20 2537 ocrdma1 hca_handle=1 hca_object=23 2538 2539HugeTLB 2540------- 2541 2542The HugeTLB controller allows to limit the HugeTLB usage per control group and 2543enforces the controller limit during page fault. 2544 2545HugeTLB Interface Files 2546~~~~~~~~~~~~~~~~~~~~~~~ 2547 2548 hugetlb.<hugepagesize>.current 2549 Show current usage for "hugepagesize" hugetlb. It exists for all 2550 the cgroup except root. 2551 2552 hugetlb.<hugepagesize>.max 2553 Set/show the hard limit of "hugepagesize" hugetlb usage. 2554 The default value is "max". It exists for all the cgroup except root. 2555 2556 hugetlb.<hugepagesize>.events 2557 A read-only flat-keyed file which exists on non-root cgroups. 2558 2559 max 2560 The number of allocation failure due to HugeTLB limit 2561 2562 hugetlb.<hugepagesize>.events.local 2563 Similar to hugetlb.<hugepagesize>.events but the fields in the file 2564 are local to the cgroup i.e. not hierarchical. The file modified event 2565 generated on this file reflects only the local events. 2566 2567 hugetlb.<hugepagesize>.numa_stat 2568 Similar to memory.numa_stat, it shows the numa information of the 2569 hugetlb pages of <hugepagesize> in this cgroup. Only active in 2570 use hugetlb pages are included. The per-node values are in bytes. 2571 2572Misc 2573---- 2574 2575The Miscellaneous cgroup provides the resource limiting and tracking 2576mechanism for the scalar resources which cannot be abstracted like the other 2577cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config 2578option. 2579 2580A resource can be added to the controller via enum misc_res_type{} in the 2581include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[] 2582in the kernel/cgroup/misc.c file. Provider of the resource must set its 2583capacity prior to using the resource by calling misc_cg_set_capacity(). 2584 2585Once a capacity is set then the resource usage can be updated using charge and 2586uncharge APIs. All of the APIs to interact with misc controller are in 2587include/linux/misc_cgroup.h. 2588 2589Misc Interface Files 2590~~~~~~~~~~~~~~~~~~~~ 2591 2592Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then: 2593 2594 misc.capacity 2595 A read-only flat-keyed file shown only in the root cgroup. It shows 2596 miscellaneous scalar resources available on the platform along with 2597 their quantities:: 2598 2599 $ cat misc.capacity 2600 res_a 50 2601 res_b 10 2602 2603 misc.current 2604 A read-only flat-keyed file shown in the all cgroups. It shows 2605 the current usage of the resources in the cgroup and its children.:: 2606 2607 $ cat misc.current 2608 res_a 3 2609 res_b 0 2610 2611 misc.max 2612 A read-write flat-keyed file shown in the non root cgroups. Allowed 2613 maximum usage of the resources in the cgroup and its children.:: 2614 2615 $ cat misc.max 2616 res_a max 2617 res_b 4 2618 2619 Limit can be set by:: 2620 2621 # echo res_a 1 > misc.max 2622 2623 Limit can be set to max by:: 2624 2625 # echo res_a max > misc.max 2626 2627 Limits can be set higher than the capacity value in the misc.capacity 2628 file. 2629 2630 misc.events 2631 A read-only flat-keyed file which exists on non-root cgroups. The 2632 following entries are defined. Unless specified otherwise, a value 2633 change in this file generates a file modified event. All fields in 2634 this file are hierarchical. 2635 2636 max 2637 The number of times the cgroup's resource usage was 2638 about to go over the max boundary. 2639 2640Migration and Ownership 2641~~~~~~~~~~~~~~~~~~~~~~~ 2642 2643A miscellaneous scalar resource is charged to the cgroup in which it is used 2644first, and stays charged to that cgroup until that resource is freed. Migrating 2645a process to a different cgroup does not move the charge to the destination 2646cgroup where the process has moved. 2647 2648Others 2649------ 2650 2651perf_event 2652~~~~~~~~~~ 2653 2654perf_event controller, if not mounted on a legacy hierarchy, is 2655automatically enabled on the v2 hierarchy so that perf events can 2656always be filtered by cgroup v2 path. The controller can still be 2657moved to a legacy hierarchy after v2 hierarchy is populated. 2658 2659 2660Non-normative information 2661------------------------- 2662 2663This section contains information that isn't considered to be a part of 2664the stable kernel API and so is subject to change. 2665 2666 2667CPU controller root cgroup process behaviour 2668~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2669 2670When distributing CPU cycles in the root cgroup each thread in this 2671cgroup is treated as if it was hosted in a separate child cgroup of the 2672root cgroup. This child cgroup weight is dependent on its thread nice 2673level. 2674 2675For details of this mapping see sched_prio_to_weight array in 2676kernel/sched/core.c file (values from this array should be scaled 2677appropriately so the neutral - nice 0 - value is 100 instead of 1024). 2678 2679 2680IO controller root cgroup process behaviour 2681~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2682 2683Root cgroup processes are hosted in an implicit leaf child node. 2684When distributing IO resources this implicit child node is taken into 2685account as if it was a normal child cgroup of the root cgroup with a 2686weight value of 200. 2687 2688 2689Namespace 2690========= 2691 2692Basics 2693------ 2694 2695cgroup namespace provides a mechanism to virtualize the view of the 2696"/proc/$PID/cgroup" file and cgroup mounts. The CLONE_NEWCGROUP clone 2697flag can be used with clone(2) and unshare(2) to create a new cgroup 2698namespace. The process running inside the cgroup namespace will have 2699its "/proc/$PID/cgroup" output restricted to cgroupns root. The 2700cgroupns root is the cgroup of the process at the time of creation of 2701the cgroup namespace. 2702 2703Without cgroup namespace, the "/proc/$PID/cgroup" file shows the 2704complete path of the cgroup of a process. In a container setup where 2705a set of cgroups and namespaces are intended to isolate processes the 2706"/proc/$PID/cgroup" file may leak potential system level information 2707to the isolated processes. For example:: 2708 2709 # cat /proc/self/cgroup 2710 0::/batchjobs/container_id1 2711 2712The path '/batchjobs/container_id1' can be considered as system-data 2713and undesirable to expose to the isolated processes. cgroup namespace 2714can be used to restrict visibility of this path. For example, before 2715creating a cgroup namespace, one would see:: 2716 2717 # ls -l /proc/self/ns/cgroup 2718 lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835] 2719 # cat /proc/self/cgroup 2720 0::/batchjobs/container_id1 2721 2722After unsharing a new namespace, the view changes:: 2723 2724 # ls -l /proc/self/ns/cgroup 2725 lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183] 2726 # cat /proc/self/cgroup 2727 0::/ 2728 2729When some thread from a multi-threaded process unshares its cgroup 2730namespace, the new cgroupns gets applied to the entire process (all 2731the threads). This is natural for the v2 hierarchy; however, for the 2732legacy hierarchies, this may be unexpected. 2733 2734A cgroup namespace is alive as long as there are processes inside or 2735mounts pinning it. When the last usage goes away, the cgroup 2736namespace is destroyed. The cgroupns root and the actual cgroups 2737remain. 2738 2739 2740The Root and Views 2741------------------ 2742 2743The 'cgroupns root' for a cgroup namespace is the cgroup in which the 2744process calling unshare(2) is running. For example, if a process in 2745/batchjobs/container_id1 cgroup calls unshare, cgroup 2746/batchjobs/container_id1 becomes the cgroupns root. For the 2747init_cgroup_ns, this is the real root ('/') cgroup. 2748 2749The cgroupns root cgroup does not change even if the namespace creator 2750process later moves to a different cgroup:: 2751 2752 # ~/unshare -c # unshare cgroupns in some cgroup 2753 # cat /proc/self/cgroup 2754 0::/ 2755 # mkdir sub_cgrp_1 2756 # echo 0 > sub_cgrp_1/cgroup.procs 2757 # cat /proc/self/cgroup 2758 0::/sub_cgrp_1 2759 2760Each process gets its namespace-specific view of "/proc/$PID/cgroup" 2761 2762Processes running inside the cgroup namespace will be able to see 2763cgroup paths (in /proc/self/cgroup) only inside their root cgroup. 2764From within an unshared cgroupns:: 2765 2766 # sleep 100000 & 2767 [1] 7353 2768 # echo 7353 > sub_cgrp_1/cgroup.procs 2769 # cat /proc/7353/cgroup 2770 0::/sub_cgrp_1 2771 2772From the initial cgroup namespace, the real cgroup path will be 2773visible:: 2774 2775 $ cat /proc/7353/cgroup 2776 0::/batchjobs/container_id1/sub_cgrp_1 2777 2778From a sibling cgroup namespace (that is, a namespace rooted at a 2779different cgroup), the cgroup path relative to its own cgroup 2780namespace root will be shown. For instance, if PID 7353's cgroup 2781namespace root is at '/batchjobs/container_id2', then it will see:: 2782 2783 # cat /proc/7353/cgroup 2784 0::/../container_id2/sub_cgrp_1 2785 2786Note that the relative path always starts with '/' to indicate that 2787its relative to the cgroup namespace root of the caller. 2788 2789 2790Migration and setns(2) 2791---------------------- 2792 2793Processes inside a cgroup namespace can move into and out of the 2794namespace root if they have proper access to external cgroups. For 2795example, from inside a namespace with cgroupns root at 2796/batchjobs/container_id1, and assuming that the global hierarchy is 2797still accessible inside cgroupns:: 2798 2799 # cat /proc/7353/cgroup 2800 0::/sub_cgrp_1 2801 # echo 7353 > batchjobs/container_id2/cgroup.procs 2802 # cat /proc/7353/cgroup 2803 0::/../container_id2 2804 2805Note that this kind of setup is not encouraged. A task inside cgroup 2806namespace should only be exposed to its own cgroupns hierarchy. 2807 2808setns(2) to another cgroup namespace is allowed when: 2809 2810(a) the process has CAP_SYS_ADMIN against its current user namespace 2811(b) the process has CAP_SYS_ADMIN against the target cgroup 2812 namespace's userns 2813 2814No implicit cgroup changes happen with attaching to another cgroup 2815namespace. It is expected that the someone moves the attaching 2816process under the target cgroup namespace root. 2817 2818 2819Interaction with Other Namespaces 2820--------------------------------- 2821 2822Namespace specific cgroup hierarchy can be mounted by a process 2823running inside a non-init cgroup namespace:: 2824 2825 # mount -t cgroup2 none $MOUNT_POINT 2826 2827This will mount the unified cgroup hierarchy with cgroupns root as the 2828filesystem root. The process needs CAP_SYS_ADMIN against its user and 2829mount namespaces. 2830 2831The virtualization of /proc/self/cgroup file combined with restricting 2832the view of cgroup hierarchy by namespace-private cgroupfs mount 2833provides a properly isolated cgroup view inside the container. 2834 2835 2836Information on Kernel Programming 2837================================= 2838 2839This section contains kernel programming information in the areas 2840where interacting with cgroup is necessary. cgroup core and 2841controllers are not covered. 2842 2843 2844Filesystem Support for Writeback 2845-------------------------------- 2846 2847A filesystem can support cgroup writeback by updating 2848address_space_operations->writepage[s]() to annotate bio's using the 2849following two functions. 2850 2851 wbc_init_bio(@wbc, @bio) 2852 Should be called for each bio carrying writeback data and 2853 associates the bio with the inode's owner cgroup and the 2854 corresponding request queue. This must be called after 2855 a queue (device) has been associated with the bio and 2856 before submission. 2857 2858 wbc_account_cgroup_owner(@wbc, @page, @bytes) 2859 Should be called for each data segment being written out. 2860 While this function doesn't care exactly when it's called 2861 during the writeback session, it's the easiest and most 2862 natural to call it as data segments are added to a bio. 2863 2864With writeback bio's annotated, cgroup support can be enabled per 2865super_block by setting SB_I_CGROUPWB in ->s_iflags. This allows for 2866selective disabling of cgroup writeback support which is helpful when 2867certain filesystem features, e.g. journaled data mode, are 2868incompatible. 2869 2870wbc_init_bio() binds the specified bio to its cgroup. Depending on 2871the configuration, the bio may be executed at a lower priority and if 2872the writeback session is holding shared resources, e.g. a journal 2873entry, may lead to priority inversion. There is no one easy solution 2874for the problem. Filesystems can try to work around specific problem 2875cases by skipping wbc_init_bio() and using bio_associate_blkg() 2876directly. 2877 2878 2879Deprecated v1 Core Features 2880=========================== 2881 2882- Multiple hierarchies including named ones are not supported. 2883 2884- All v1 mount options are not supported. 2885 2886- The "tasks" file is removed and "cgroup.procs" is not sorted. 2887 2888- "cgroup.clone_children" is removed. 2889 2890- /proc/cgroups is meaningless for v2. Use "cgroup.controllers" file 2891 at the root instead. 2892 2893 2894Issues with v1 and Rationales for v2 2895==================================== 2896 2897Multiple Hierarchies 2898-------------------- 2899 2900cgroup v1 allowed an arbitrary number of hierarchies and each 2901hierarchy could host any number of controllers. While this seemed to 2902provide a high level of flexibility, it wasn't useful in practice. 2903 2904For example, as there is only one instance of each controller, utility 2905type controllers such as freezer which can be useful in all 2906hierarchies could only be used in one. The issue is exacerbated by 2907the fact that controllers couldn't be moved to another hierarchy once 2908hierarchies were populated. Another issue was that all controllers 2909bound to a hierarchy were forced to have exactly the same view of the 2910hierarchy. It wasn't possible to vary the granularity depending on 2911the specific controller. 2912 2913In practice, these issues heavily limited which controllers could be 2914put on the same hierarchy and most configurations resorted to putting 2915each controller on its own hierarchy. Only closely related ones, such 2916as the cpu and cpuacct controllers, made sense to be put on the same 2917hierarchy. This often meant that userland ended up managing multiple 2918similar hierarchies repeating the same steps on each hierarchy 2919whenever a hierarchy management operation was necessary. 2920 2921Furthermore, support for multiple hierarchies came at a steep cost. 2922It greatly complicated cgroup core implementation but more importantly 2923the support for multiple hierarchies restricted how cgroup could be 2924used in general and what controllers was able to do. 2925 2926There was no limit on how many hierarchies there might be, which meant 2927that a thread's cgroup membership couldn't be described in finite 2928length. The key might contain any number of entries and was unlimited 2929in length, which made it highly awkward to manipulate and led to 2930addition of controllers which existed only to identify membership, 2931which in turn exacerbated the original problem of proliferating number 2932of hierarchies. 2933 2934Also, as a controller couldn't have any expectation regarding the 2935topologies of hierarchies other controllers might be on, each 2936controller had to assume that all other controllers were attached to 2937completely orthogonal hierarchies. This made it impossible, or at 2938least very cumbersome, for controllers to cooperate with each other. 2939 2940In most use cases, putting controllers on hierarchies which are 2941completely orthogonal to each other isn't necessary. What usually is 2942called for is the ability to have differing levels of granularity 2943depending on the specific controller. In other words, hierarchy may 2944be collapsed from leaf towards root when viewed from specific 2945controllers. For example, a given configuration might not care about 2946how memory is distributed beyond a certain level while still wanting 2947to control how CPU cycles are distributed. 2948 2949 2950Thread Granularity 2951------------------ 2952 2953cgroup v1 allowed threads of a process to belong to different cgroups. 2954This didn't make sense for some controllers and those controllers 2955ended up implementing different ways to ignore such situations but 2956much more importantly it blurred the line between API exposed to 2957individual applications and system management interface. 2958 2959Generally, in-process knowledge is available only to the process 2960itself; thus, unlike service-level organization of processes, 2961categorizing threads of a process requires active participation from 2962the application which owns the target process. 2963 2964cgroup v1 had an ambiguously defined delegation model which got abused 2965in combination with thread granularity. cgroups were delegated to 2966individual applications so that they can create and manage their own 2967sub-hierarchies and control resource distributions along them. This 2968effectively raised cgroup to the status of a syscall-like API exposed 2969to lay programs. 2970 2971First of all, cgroup has a fundamentally inadequate interface to be 2972exposed this way. For a process to access its own knobs, it has to 2973extract the path on the target hierarchy from /proc/self/cgroup, 2974construct the path by appending the name of the knob to the path, open 2975and then read and/or write to it. This is not only extremely clunky 2976and unusual but also inherently racy. There is no conventional way to 2977define transaction across the required steps and nothing can guarantee 2978that the process would actually be operating on its own sub-hierarchy. 2979 2980cgroup controllers implemented a number of knobs which would never be 2981accepted as public APIs because they were just adding control knobs to 2982system-management pseudo filesystem. cgroup ended up with interface 2983knobs which were not properly abstracted or refined and directly 2984revealed kernel internal details. These knobs got exposed to 2985individual applications through the ill-defined delegation mechanism 2986effectively abusing cgroup as a shortcut to implementing public APIs 2987without going through the required scrutiny. 2988 2989This was painful for both userland and kernel. Userland ended up with 2990misbehaving and poorly abstracted interfaces and kernel exposing and 2991locked into constructs inadvertently. 2992 2993 2994Competition Between Inner Nodes and Threads 2995------------------------------------------- 2996 2997cgroup v1 allowed threads to be in any cgroups which created an 2998interesting problem where threads belonging to a parent cgroup and its 2999children cgroups competed for resources. This was nasty as two 3000different types of entities competed and there was no obvious way to 3001settle it. Different controllers did different things. 3002 3003The cpu controller considered threads and cgroups as equivalents and 3004mapped nice levels to cgroup weights. This worked for some cases but 3005fell flat when children wanted to be allocated specific ratios of CPU 3006cycles and the number of internal threads fluctuated - the ratios 3007constantly changed as the number of competing entities fluctuated. 3008There also were other issues. The mapping from nice level to weight 3009wasn't obvious or universal, and there were various other knobs which 3010simply weren't available for threads. 3011 3012The io controller implicitly created a hidden leaf node for each 3013cgroup to host the threads. The hidden leaf had its own copies of all 3014the knobs with ``leaf_`` prefixed. While this allowed equivalent 3015control over internal threads, it was with serious drawbacks. It 3016always added an extra layer of nesting which wouldn't be necessary 3017otherwise, made the interface messy and significantly complicated the 3018implementation. 3019 3020The memory controller didn't have a way to control what happened 3021between internal tasks and child cgroups and the behavior was not 3022clearly defined. There were attempts to add ad-hoc behaviors and 3023knobs to tailor the behavior to specific workloads which would have 3024led to problems extremely difficult to resolve in the long term. 3025 3026Multiple controllers struggled with internal tasks and came up with 3027different ways to deal with it; unfortunately, all the approaches were 3028severely flawed and, furthermore, the widely different behaviors 3029made cgroup as a whole highly inconsistent. 3030 3031This clearly is a problem which needs to be addressed from cgroup core 3032in a uniform way. 3033 3034 3035Other Interface Issues 3036---------------------- 3037 3038cgroup v1 grew without oversight and developed a large number of 3039idiosyncrasies and inconsistencies. One issue on the cgroup core side 3040was how an empty cgroup was notified - a userland helper binary was 3041forked and executed for each event. The event delivery wasn't 3042recursive or delegatable. The limitations of the mechanism also led 3043to in-kernel event delivery filtering mechanism further complicating 3044the interface. 3045 3046Controller interfaces were problematic too. An extreme example is 3047controllers completely ignoring hierarchical organization and treating 3048all cgroups as if they were all located directly under the root 3049cgroup. Some controllers exposed a large amount of inconsistent 3050implementation details to userland. 3051 3052There also was no consistency across controllers. When a new cgroup 3053was created, some controllers defaulted to not imposing extra 3054restrictions while others disallowed any resource usage until 3055explicitly configured. Configuration knobs for the same type of 3056control used widely differing naming schemes and formats. Statistics 3057and information knobs were named arbitrarily and used different 3058formats and units even in the same controller. 3059 3060cgroup v2 establishes common conventions where appropriate and updates 3061controllers so that they expose minimal and consistent interfaces. 3062 3063 3064Controller Issues and Remedies 3065------------------------------ 3066 3067Memory 3068~~~~~~ 3069 3070The original lower boundary, the soft limit, is defined as a limit 3071that is per default unset. As a result, the set of cgroups that 3072global reclaim prefers is opt-in, rather than opt-out. The costs for 3073optimizing these mostly negative lookups are so high that the 3074implementation, despite its enormous size, does not even provide the 3075basic desirable behavior. First off, the soft limit has no 3076hierarchical meaning. All configured groups are organized in a global 3077rbtree and treated like equal peers, regardless where they are located 3078in the hierarchy. This makes subtree delegation impossible. Second, 3079the soft limit reclaim pass is so aggressive that it not just 3080introduces high allocation latencies into the system, but also impacts 3081system performance due to overreclaim, to the point where the feature 3082becomes self-defeating. 3083 3084The memory.low boundary on the other hand is a top-down allocated 3085reserve. A cgroup enjoys reclaim protection when it's within its 3086effective low, which makes delegation of subtrees possible. It also 3087enjoys having reclaim pressure proportional to its overage when 3088above its effective low. 3089 3090The original high boundary, the hard limit, is defined as a strict 3091limit that can not budge, even if the OOM killer has to be called. 3092But this generally goes against the goal of making the most out of the 3093available memory. The memory consumption of workloads varies during 3094runtime, and that requires users to overcommit. But doing that with a 3095strict upper limit requires either a fairly accurate prediction of the 3096working set size or adding slack to the limit. Since working set size 3097estimation is hard and error prone, and getting it wrong results in 3098OOM kills, most users tend to err on the side of a looser limit and 3099end up wasting precious resources. 3100 3101The memory.high boundary on the other hand can be set much more 3102conservatively. When hit, it throttles allocations by forcing them 3103into direct reclaim to work off the excess, but it never invokes the 3104OOM killer. As a result, a high boundary that is chosen too 3105aggressively will not terminate the processes, but instead it will 3106lead to gradual performance degradation. The user can monitor this 3107and make corrections until the minimal memory footprint that still 3108gives acceptable performance is found. 3109 3110In extreme cases, with many concurrent allocations and a complete 3111breakdown of reclaim progress within the group, the high boundary can 3112be exceeded. But even then it's mostly better to satisfy the 3113allocation from the slack available in other groups or the rest of the 3114system than killing the group. Otherwise, memory.max is there to 3115limit this type of spillover and ultimately contain buggy or even 3116malicious applications. 3117 3118Setting the original memory.limit_in_bytes below the current usage was 3119subject to a race condition, where concurrent charges could cause the 3120limit setting to fail. memory.max on the other hand will first set the 3121limit to prevent new charges, and then reclaim and OOM kill until the 3122new limit is met - or the task writing to memory.max is killed. 3123 3124The combined memory+swap accounting and limiting is replaced by real 3125control over swap space. 3126 3127The main argument for a combined memory+swap facility in the original 3128cgroup design was that global or parental pressure would always be 3129able to swap all anonymous memory of a child group, regardless of the 3130child's own (possibly untrusted) configuration. However, untrusted 3131groups can sabotage swapping by other means - such as referencing its 3132anonymous memory in a tight loop - and an admin can not assume full 3133swappability when overcommitting untrusted jobs. 3134 3135For trusted jobs, on the other hand, a combined counter is not an 3136intuitive userspace interface, and it flies in the face of the idea 3137that cgroup controllers should account and limit specific physical 3138resources. Swap space is a resource like all others in the system, 3139and that's why unified hierarchy allows distributing it separately. 3140