Based on kernel version 4.7.2. Page generated on 2016-08-22 22:48 EST.
1 Documentation for /proc/sys/vm/* kernel version 2.6.29 2 (c) 1998, 1999, Rik van Riel <firstname.lastname@example.org> 3 (c) 2008 Peter W. Morreale <email@example.com> 4 5 For general info and legal blurb, please look in README. 6 7 ============================================================== 8 9 This file contains the documentation for the sysctl files in 10 /proc/sys/vm and is valid for Linux kernel version 2.6.29. 11 12 The files in this directory can be used to tune the operation 13 of the virtual memory (VM) subsystem of the Linux kernel and 14 the writeout of dirty data to disk. 15 16 Default values and initialization routines for most of these 17 files can be found in mm/swap.c. 18 19 Currently, these files are in /proc/sys/vm: 20 21 - admin_reserve_kbytes 22 - block_dump 23 - compact_memory 24 - compact_unevictable_allowed 25 - dirty_background_bytes 26 - dirty_background_ratio 27 - dirty_bytes 28 - dirty_expire_centisecs 29 - dirty_ratio 30 - dirty_writeback_centisecs 31 - drop_caches 32 - extfrag_threshold 33 - hugepages_treat_as_movable 34 - hugetlb_shm_group 35 - laptop_mode 36 - legacy_va_layout 37 - lowmem_reserve_ratio 38 - max_map_count 39 - memory_failure_early_kill 40 - memory_failure_recovery 41 - min_free_kbytes 42 - min_slab_ratio 43 - min_unmapped_ratio 44 - mmap_min_addr 45 - mmap_rnd_bits 46 - mmap_rnd_compat_bits 47 - nr_hugepages 48 - nr_overcommit_hugepages 49 - nr_trim_pages (only if CONFIG_MMU=n) 50 - numa_zonelist_order 51 - oom_dump_tasks 52 - oom_kill_allocating_task 53 - overcommit_kbytes 54 - overcommit_memory 55 - overcommit_ratio 56 - page-cluster 57 - panic_on_oom 58 - percpu_pagelist_fraction 59 - stat_interval 60 - stat_refresh 61 - swappiness 62 - user_reserve_kbytes 63 - vfs_cache_pressure 64 - zone_reclaim_mode 65 66 ============================================================== 67 68 admin_reserve_kbytes 69 70 The amount of free memory in the system that should be reserved for users 71 with the capability cap_sys_admin. 72 73 admin_reserve_kbytes defaults to min(3% of free pages, 8MB) 74 75 That should provide enough for the admin to log in and kill a process, 76 if necessary, under the default overcommit 'guess' mode. 77 78 Systems running under overcommit 'never' should increase this to account 79 for the full Virtual Memory Size of programs used to recover. Otherwise, 80 root may not be able to log in to recover the system. 81 82 How do you calculate a minimum useful reserve? 83 84 sshd or login + bash (or some other shell) + top (or ps, kill, etc.) 85 86 For overcommit 'guess', we can sum resident set sizes (RSS). 87 On x86_64 this is about 8MB. 88 89 For overcommit 'never', we can take the max of their virtual sizes (VSZ) 90 and add the sum of their RSS. 91 On x86_64 this is about 128MB. 92 93 Changing this takes effect whenever an application requests memory. 94 95 ============================================================== 96 97 block_dump 98 99 block_dump enables block I/O debugging when set to a nonzero value. More 100 information on block I/O debugging is in Documentation/laptops/laptop-mode.txt. 101 102 ============================================================== 103 104 compact_memory 105 106 Available only when CONFIG_COMPACTION is set. When 1 is written to the file, 107 all zones are compacted such that free memory is available in contiguous 108 blocks where possible. This can be important for example in the allocation of 109 huge pages although processes will also directly compact memory as required. 110 111 ============================================================== 112 113 compact_unevictable_allowed 114 115 Available only when CONFIG_COMPACTION is set. When set to 1, compaction is 116 allowed to examine the unevictable lru (mlocked pages) for pages to compact. 117 This should be used on systems where stalls for minor page faults are an 118 acceptable trade for large contiguous free memory. Set to 0 to prevent 119 compaction from moving pages that are unevictable. Default value is 1. 120 121 ============================================================== 122 123 dirty_background_bytes 124 125 Contains the amount of dirty memory at which the background kernel 126 flusher threads will start writeback. 127 128 Note: dirty_background_bytes is the counterpart of dirty_background_ratio. Only 129 one of them may be specified at a time. When one sysctl is written it is 130 immediately taken into account to evaluate the dirty memory limits and the 131 other appears as 0 when read. 132 133 ============================================================== 134 135 dirty_background_ratio 136 137 Contains, as a percentage of total available memory that contains free pages 138 and reclaimable pages, the number of pages at which the background kernel 139 flusher threads will start writing out dirty data. 140 141 The total available memory is not equal to total system memory. 142 143 ============================================================== 144 145 dirty_bytes 146 147 Contains the amount of dirty memory at which a process generating disk writes 148 will itself start writeback. 149 150 Note: dirty_bytes is the counterpart of dirty_ratio. Only one of them may be 151 specified at a time. When one sysctl is written it is immediately taken into 152 account to evaluate the dirty memory limits and the other appears as 0 when 153 read. 154 155 Note: the minimum value allowed for dirty_bytes is two pages (in bytes); any 156 value lower than this limit will be ignored and the old configuration will be 157 retained. 158 159 ============================================================== 160 161 dirty_expire_centisecs 162 163 This tunable is used to define when dirty data is old enough to be eligible 164 for writeout by the kernel flusher threads. It is expressed in 100'ths 165 of a second. Data which has been dirty in-memory for longer than this 166 interval will be written out next time a flusher thread wakes up. 167 168 ============================================================== 169 170 dirty_ratio 171 172 Contains, as a percentage of total available memory that contains free pages 173 and reclaimable pages, the number of pages at which a process which is 174 generating disk writes will itself start writing out dirty data. 175 176 The total available memory is not equal to total system memory. 177 178 ============================================================== 179 180 dirty_writeback_centisecs 181 182 The kernel flusher threads will periodically wake up and write `old' data 183 out to disk. This tunable expresses the interval between those wakeups, in 184 100'ths of a second. 185 186 Setting this to zero disables periodic writeback altogether. 187 188 ============================================================== 189 190 drop_caches 191 192 Writing to this will cause the kernel to drop clean caches, as well as 193 reclaimable slab objects like dentries and inodes. Once dropped, their 194 memory becomes free. 195 196 To free pagecache: 197 echo 1 > /proc/sys/vm/drop_caches 198 To free reclaimable slab objects (includes dentries and inodes): 199 echo 2 > /proc/sys/vm/drop_caches 200 To free slab objects and pagecache: 201 echo 3 > /proc/sys/vm/drop_caches 202 203 This is a non-destructive operation and will not free any dirty objects. 204 To increase the number of objects freed by this operation, the user may run 205 `sync' prior to writing to /proc/sys/vm/drop_caches. This will minimize the 206 number of dirty objects on the system and create more candidates to be 207 dropped. 208 209 This file is not a means to control the growth of the various kernel caches 210 (inodes, dentries, pagecache, etc...) These objects are automatically 211 reclaimed by the kernel when memory is needed elsewhere on the system. 212 213 Use of this file can cause performance problems. Since it discards cached 214 objects, it may cost a significant amount of I/O and CPU to recreate the 215 dropped objects, especially if they were under heavy use. Because of this, 216 use outside of a testing or debugging environment is not recommended. 217 218 You may see informational messages in your kernel log when this file is 219 used: 220 221 cat (1234): drop_caches: 3 222 223 These are informational only. They do not mean that anything is wrong 224 with your system. To disable them, echo 4 (bit 3) into drop_caches. 225 226 ============================================================== 227 228 extfrag_threshold 229 230 This parameter affects whether the kernel will compact memory or direct 231 reclaim to satisfy a high-order allocation. The extfrag/extfrag_index file in 232 debugfs shows what the fragmentation index for each order is in each zone in 233 the system. Values tending towards 0 imply allocations would fail due to lack 234 of memory, values towards 1000 imply failures are due to fragmentation and -1 235 implies that the allocation will succeed as long as watermarks are met. 236 237 The kernel will not compact memory in a zone if the 238 fragmentation index is <= extfrag_threshold. The default value is 500. 239 240 ============================================================== 241 242 hugepages_treat_as_movable 243 244 This parameter controls whether we can allocate hugepages from ZONE_MOVABLE 245 or not. If set to non-zero, hugepages can be allocated from ZONE_MOVABLE. 246 ZONE_MOVABLE is created when kernel boot parameter kernelcore= is specified, 247 so this parameter has no effect if used without kernelcore=. 248 249 Hugepage migration is now available in some situations which depend on the 250 architecture and/or the hugepage size. If a hugepage supports migration, 251 allocation from ZONE_MOVABLE is always enabled for the hugepage regardless 252 of the value of this parameter. 253 IOW, this parameter affects only non-migratable hugepages. 254 255 Assuming that hugepages are not migratable in your system, one usecase of 256 this parameter is that users can make hugepage pool more extensible by 257 enabling the allocation from ZONE_MOVABLE. This is because on ZONE_MOVABLE 258 page reclaim/migration/compaction work more and you can get contiguous 259 memory more likely. Note that using ZONE_MOVABLE for non-migratable 260 hugepages can do harm to other features like memory hotremove (because 261 memory hotremove expects that memory blocks on ZONE_MOVABLE are always 262 removable,) so it's a trade-off responsible for the users. 263 264 ============================================================== 265 266 hugetlb_shm_group 267 268 hugetlb_shm_group contains group id that is allowed to create SysV 269 shared memory segment using hugetlb page. 270 271 ============================================================== 272 273 laptop_mode 274 275 laptop_mode is a knob that controls "laptop mode". All the things that are 276 controlled by this knob are discussed in Documentation/laptops/laptop-mode.txt. 277 278 ============================================================== 279 280 legacy_va_layout 281 282 If non-zero, this sysctl disables the new 32-bit mmap layout - the kernel 283 will use the legacy (2.4) layout for all processes. 284 285 ============================================================== 286 287 lowmem_reserve_ratio 288 289 For some specialised workloads on highmem machines it is dangerous for 290 the kernel to allow process memory to be allocated from the "lowmem" 291 zone. This is because that memory could then be pinned via the mlock() 292 system call, or by unavailability of swapspace. 293 294 And on large highmem machines this lack of reclaimable lowmem memory 295 can be fatal. 296 297 So the Linux page allocator has a mechanism which prevents allocations 298 which _could_ use highmem from using too much lowmem. This means that 299 a certain amount of lowmem is defended from the possibility of being 300 captured into pinned user memory. 301 302 (The same argument applies to the old 16 megabyte ISA DMA region. This 303 mechanism will also defend that region from allocations which could use 304 highmem or lowmem). 305 306 The `lowmem_reserve_ratio' tunable determines how aggressive the kernel is 307 in defending these lower zones. 308 309 If you have a machine which uses highmem or ISA DMA and your 310 applications are using mlock(), or if you are running with no swap then 311 you probably should change the lowmem_reserve_ratio setting. 312 313 The lowmem_reserve_ratio is an array. You can see them by reading this file. 314 - 315 % cat /proc/sys/vm/lowmem_reserve_ratio 316 256 256 32 317 - 318 Note: # of this elements is one fewer than number of zones. Because the highest 319 zone's value is not necessary for following calculation. 320 321 But, these values are not used directly. The kernel calculates # of protection 322 pages for each zones from them. These are shown as array of protection pages 323 in /proc/zoneinfo like followings. (This is an example of x86-64 box). 324 Each zone has an array of protection pages like this. 325 326 - 327 Node 0, zone DMA 328 pages free 1355 329 min 3 330 low 3 331 high 4 332 : 333 : 334 numa_other 0 335 protection: (0, 2004, 2004, 2004) 336 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 337 pagesets 338 cpu: 0 pcp: 0 339 : 340 - 341 These protections are added to score to judge whether this zone should be used 342 for page allocation or should be reclaimed. 343 344 In this example, if normal pages (index=2) are required to this DMA zone and 345 watermark[WMARK_HIGH] is used for watermark, the kernel judges this zone should 346 not be used because pages_free(1355) is smaller than watermark + protection 347 (4 + 2004 = 2008). If this protection value is 0, this zone would be used for 348 normal page requirement. If requirement is DMA zone(index=0), protection 349 (=0) is used. 350 351 zone[i]'s protection[j] is calculated by following expression. 352 353 (i < j): 354 zone[i]->protection[j] 355 = (total sums of managed_pages from zone[i+1] to zone[j] on the node) 356 / lowmem_reserve_ratio[i]; 357 (i = j): 358 (should not be protected. = 0; 359 (i > j): 360 (not necessary, but looks 0) 361 362 The default values of lowmem_reserve_ratio[i] are 363 256 (if zone[i] means DMA or DMA32 zone) 364 32 (others). 365 As above expression, they are reciprocal number of ratio. 366 256 means 1/256. # of protection pages becomes about "0.39%" of total managed 367 pages of higher zones on the node. 368 369 If you would like to protect more pages, smaller values are effective. 370 The minimum value is 1 (1/1 -> 100%). 371 372 ============================================================== 373 374 max_map_count: 375 376 This file contains the maximum number of memory map areas a process 377 may have. Memory map areas are used as a side-effect of calling 378 malloc, directly by mmap and mprotect, and also when loading shared 379 libraries. 380 381 While most applications need less than a thousand maps, certain 382 programs, particularly malloc debuggers, may consume lots of them, 383 e.g., up to one or two maps per allocation. 384 385 The default value is 65536. 386 387 ============================================================= 388 389 memory_failure_early_kill: 390 391 Control how to kill processes when uncorrected memory error (typically 392 a 2bit error in a memory module) is detected in the background by hardware 393 that cannot be handled by the kernel. In some cases (like the page 394 still having a valid copy on disk) the kernel will handle the failure 395 transparently without affecting any applications. But if there is 396 no other uptodate copy of the data it will kill to prevent any data 397 corruptions from propagating. 398 399 1: Kill all processes that have the corrupted and not reloadable page mapped 400 as soon as the corruption is detected. Note this is not supported 401 for a few types of pages, like kernel internally allocated data or 402 the swap cache, but works for the majority of user pages. 403 404 0: Only unmap the corrupted page from all processes and only kill a process 405 who tries to access it. 406 407 The kill is done using a catchable SIGBUS with BUS_MCEERR_AO, so processes can 408 handle this if they want to. 409 410 This is only active on architectures/platforms with advanced machine 411 check handling and depends on the hardware capabilities. 412 413 Applications can override this setting individually with the PR_MCE_KILL prctl 414 415 ============================================================== 416 417 memory_failure_recovery 418 419 Enable memory failure recovery (when supported by the platform) 420 421 1: Attempt recovery. 422 423 0: Always panic on a memory failure. 424 425 ============================================================== 426 427 min_free_kbytes: 428 429 This is used to force the Linux VM to keep a minimum number 430 of kilobytes free. The VM uses this number to compute a 431 watermark[WMARK_MIN] value for each lowmem zone in the system. 432 Each lowmem zone gets a number of reserved free pages based 433 proportionally on its size. 434 435 Some minimal amount of memory is needed to satisfy PF_MEMALLOC 436 allocations; if you set this to lower than 1024KB, your system will 437 become subtly broken, and prone to deadlock under high loads. 438 439 Setting this too high will OOM your machine instantly. 440 441 ============================================================= 442 443 min_slab_ratio: 444 445 This is available only on NUMA kernels. 446 447 A percentage of the total pages in each zone. On Zone reclaim 448 (fallback from the local zone occurs) slabs will be reclaimed if more 449 than this percentage of pages in a zone are reclaimable slab pages. 450 This insures that the slab growth stays under control even in NUMA 451 systems that rarely perform global reclaim. 452 453 The default is 5 percent. 454 455 Note that slab reclaim is triggered in a per zone / node fashion. 456 The process of reclaiming slab memory is currently not node specific 457 and may not be fast. 458 459 ============================================================= 460 461 min_unmapped_ratio: 462 463 This is available only on NUMA kernels. 464 465 This is a percentage of the total pages in each zone. Zone reclaim will 466 only occur if more than this percentage of pages are in a state that 467 zone_reclaim_mode allows to be reclaimed. 468 469 If zone_reclaim_mode has the value 4 OR'd, then the percentage is compared 470 against all file-backed unmapped pages including swapcache pages and tmpfs 471 files. Otherwise, only unmapped pages backed by normal files but not tmpfs 472 files and similar are considered. 473 474 The default is 1 percent. 475 476 ============================================================== 477 478 mmap_min_addr 479 480 This file indicates the amount of address space which a user process will 481 be restricted from mmapping. Since kernel null dereference bugs could 482 accidentally operate based on the information in the first couple of pages 483 of memory userspace processes should not be allowed to write to them. By 484 default this value is set to 0 and no protections will be enforced by the 485 security module. Setting this value to something like 64k will allow the 486 vast majority of applications to work correctly and provide defense in depth 487 against future potential kernel bugs. 488 489 ============================================================== 490 491 mmap_rnd_bits: 492 493 This value can be used to select the number of bits to use to 494 determine the random offset to the base address of vma regions 495 resulting from mmap allocations on architectures which support 496 tuning address space randomization. This value will be bounded 497 by the architecture's minimum and maximum supported values. 498 499 This value can be changed after boot using the 500 /proc/sys/vm/mmap_rnd_bits tunable 501 502 ============================================================== 503 504 mmap_rnd_compat_bits: 505 506 This value can be used to select the number of bits to use to 507 determine the random offset to the base address of vma regions 508 resulting from mmap allocations for applications run in 509 compatibility mode on architectures which support tuning address 510 space randomization. This value will be bounded by the 511 architecture's minimum and maximum supported values. 512 513 This value can be changed after boot using the 514 /proc/sys/vm/mmap_rnd_compat_bits tunable 515 516 ============================================================== 517 518 nr_hugepages 519 520 Change the minimum size of the hugepage pool. 521 522 See Documentation/vm/hugetlbpage.txt 523 524 ============================================================== 525 526 nr_overcommit_hugepages 527 528 Change the maximum size of the hugepage pool. The maximum is 529 nr_hugepages + nr_overcommit_hugepages. 530 531 See Documentation/vm/hugetlbpage.txt 532 533 ============================================================== 534 535 nr_trim_pages 536 537 This is available only on NOMMU kernels. 538 539 This value adjusts the excess page trimming behaviour of power-of-2 aligned 540 NOMMU mmap allocations. 541 542 A value of 0 disables trimming of allocations entirely, while a value of 1 543 trims excess pages aggressively. Any value >= 1 acts as the watermark where 544 trimming of allocations is initiated. 545 546 The default value is 1. 547 548 See Documentation/nommu-mmap.txt for more information. 549 550 ============================================================== 551 552 numa_zonelist_order 553 554 This sysctl is only for NUMA. 555 'where the memory is allocated from' is controlled by zonelists. 556 (This documentation ignores ZONE_HIGHMEM/ZONE_DMA32 for simple explanation. 557 you may be able to read ZONE_DMA as ZONE_DMA32...) 558 559 In non-NUMA case, a zonelist for GFP_KERNEL is ordered as following. 560 ZONE_NORMAL -> ZONE_DMA 561 This means that a memory allocation request for GFP_KERNEL will 562 get memory from ZONE_DMA only when ZONE_NORMAL is not available. 563 564 In NUMA case, you can think of following 2 types of order. 565 Assume 2 node NUMA and below is zonelist of Node(0)'s GFP_KERNEL 566 567 (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA -> Node(1) ZONE_NORMAL 568 (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORMAL -> Node(0) ZONE_DMA. 569 570 Type(A) offers the best locality for processes on Node(0), but ZONE_DMA 571 will be used before ZONE_NORMAL exhaustion. This increases possibility of 572 out-of-memory(OOM) of ZONE_DMA because ZONE_DMA is tend to be small. 573 574 Type(B) cannot offer the best locality but is more robust against OOM of 575 the DMA zone. 576 577 Type(A) is called as "Node" order. Type (B) is "Zone" order. 578 579 "Node order" orders the zonelists by node, then by zone within each node. 580 Specify "[Nn]ode" for node order 581 582 "Zone Order" orders the zonelists by zone type, then by node within each 583 zone. Specify "[Zz]one" for zone order. 584 585 Specify "[Dd]efault" to request automatic configuration. 586 587 On 32-bit, the Normal zone needs to be preserved for allocations accessible 588 by the kernel, so "zone" order will be selected. 589 590 On 64-bit, devices that require DMA32/DMA are relatively rare, so "node" 591 order will be selected. 592 593 Default order is recommended unless this is causing problems for your 594 system/application. 595 596 ============================================================== 597 598 oom_dump_tasks 599 600 Enables a system-wide task dump (excluding kernel threads) to be produced 601 when the kernel performs an OOM-killing and includes such information as 602 pid, uid, tgid, vm size, rss, nr_ptes, nr_pmds, swapents, oom_score_adj 603 score, and name. This is helpful to determine why the OOM killer was 604 invoked, to identify the rogue task that caused it, and to determine why 605 the OOM killer chose the task it did to kill. 606 607 If this is set to zero, this information is suppressed. On very 608 large systems with thousands of tasks it may not be feasible to dump 609 the memory state information for each one. Such systems should not 610 be forced to incur a performance penalty in OOM conditions when the 611 information may not be desired. 612 613 If this is set to non-zero, this information is shown whenever the 614 OOM killer actually kills a memory-hogging task. 615 616 The default value is 1 (enabled). 617 618 ============================================================== 619 620 oom_kill_allocating_task 621 622 This enables or disables killing the OOM-triggering task in 623 out-of-memory situations. 624 625 If this is set to zero, the OOM killer will scan through the entire 626 tasklist and select a task based on heuristics to kill. This normally 627 selects a rogue memory-hogging task that frees up a large amount of 628 memory when killed. 629 630 If this is set to non-zero, the OOM killer simply kills the task that 631 triggered the out-of-memory condition. This avoids the expensive 632 tasklist scan. 633 634 If panic_on_oom is selected, it takes precedence over whatever value 635 is used in oom_kill_allocating_task. 636 637 The default value is 0. 638 639 ============================================================== 640 641 overcommit_kbytes: 642 643 When overcommit_memory is set to 2, the committed address space is not 644 permitted to exceed swap plus this amount of physical RAM. See below. 645 646 Note: overcommit_kbytes is the counterpart of overcommit_ratio. Only one 647 of them may be specified at a time. Setting one disables the other (which 648 then appears as 0 when read). 649 650 ============================================================== 651 652 overcommit_memory: 653 654 This value contains a flag that enables memory overcommitment. 655 656 When this flag is 0, the kernel attempts to estimate the amount 657 of free memory left when userspace requests more memory. 658 659 When this flag is 1, the kernel pretends there is always enough 660 memory until it actually runs out. 661 662 When this flag is 2, the kernel uses a "never overcommit" 663 policy that attempts to prevent any overcommit of memory. 664 Note that user_reserve_kbytes affects this policy. 665 666 This feature can be very useful because there are a lot of 667 programs that malloc() huge amounts of memory "just-in-case" 668 and don't use much of it. 669 670 The default value is 0. 671 672 See Documentation/vm/overcommit-accounting and 673 mm/mmap.c::__vm_enough_memory() for more information. 674 675 ============================================================== 676 677 overcommit_ratio: 678 679 When overcommit_memory is set to 2, the committed address 680 space is not permitted to exceed swap plus this percentage 681 of physical RAM. See above. 682 683 ============================================================== 684 685 page-cluster 686 687 page-cluster controls the number of pages up to which consecutive pages 688 are read in from swap in a single attempt. This is the swap counterpart 689 to page cache readahead. 690 The mentioned consecutivity is not in terms of virtual/physical addresses, 691 but consecutive on swap space - that means they were swapped out together. 692 693 It is a logarithmic value - setting it to zero means "1 page", setting 694 it to 1 means "2 pages", setting it to 2 means "4 pages", etc. 695 Zero disables swap readahead completely. 696 697 The default value is three (eight pages at a time). There may be some 698 small benefits in tuning this to a different value if your workload is 699 swap-intensive. 700 701 Lower values mean lower latencies for initial faults, but at the same time 702 extra faults and I/O delays for following faults if they would have been part of 703 that consecutive pages readahead would have brought in. 704 705 ============================================================= 706 707 panic_on_oom 708 709 This enables or disables panic on out-of-memory feature. 710 711 If this is set to 0, the kernel will kill some rogue process, 712 called oom_killer. Usually, oom_killer can kill rogue processes and 713 system will survive. 714 715 If this is set to 1, the kernel panics when out-of-memory happens. 716 However, if a process limits using nodes by mempolicy/cpusets, 717 and those nodes become memory exhaustion status, one process 718 may be killed by oom-killer. No panic occurs in this case. 719 Because other nodes' memory may be free. This means system total status 720 may be not fatal yet. 721 722 If this is set to 2, the kernel panics compulsorily even on the 723 above-mentioned. Even oom happens under memory cgroup, the whole 724 system panics. 725 726 The default value is 0. 727 1 and 2 are for failover of clustering. Please select either 728 according to your policy of failover. 729 panic_on_oom=2+kdump gives you very strong tool to investigate 730 why oom happens. You can get snapshot. 731 732 ============================================================= 733 734 percpu_pagelist_fraction 735 736 This is the fraction of pages at most (high mark pcp->high) in each zone that 737 are allocated for each per cpu page list. The min value for this is 8. It 738 means that we don't allow more than 1/8th of pages in each zone to be 739 allocated in any single per_cpu_pagelist. This entry only changes the value 740 of hot per cpu pagelists. User can specify a number like 100 to allocate 741 1/100th of each zone to each per cpu page list. 742 743 The batch value of each per cpu pagelist is also updated as a result. It is 744 set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8) 745 746 The initial value is zero. Kernel does not use this value at boot time to set 747 the high water marks for each per cpu page list. If the user writes '0' to this 748 sysctl, it will revert to this default behavior. 749 750 ============================================================== 751 752 stat_interval 753 754 The time interval between which vm statistics are updated. The default 755 is 1 second. 756 757 ============================================================== 758 759 stat_refresh 760 761 Any read or write (by root only) flushes all the per-cpu vm statistics 762 into their global totals, for more accurate reports when testing 763 e.g. cat /proc/sys/vm/stat_refresh /proc/meminfo 764 765 As a side-effect, it also checks for negative totals (elsewhere reported 766 as 0) and "fails" with EINVAL if any are found, with a warning in dmesg. 767 (At time of writing, a few stats are known sometimes to be found negative, 768 with no ill effects: errors and warnings on these stats are suppressed.) 769 770 ============================================================== 771 772 swappiness 773 774 This control is used to define how aggressive the kernel will swap 775 memory pages. Higher values will increase agressiveness, lower values 776 decrease the amount of swap. A value of 0 instructs the kernel not to 777 initiate swap until the amount of free and file-backed pages is less 778 than the high water mark in a zone. 779 780 The default value is 60. 781 782 ============================================================== 783 784 - user_reserve_kbytes 785 786 When overcommit_memory is set to 2, "never overcommit" mode, reserve 787 min(3% of current process size, user_reserve_kbytes) of free memory. 788 This is intended to prevent a user from starting a single memory hogging 789 process, such that they cannot recover (kill the hog). 790 791 user_reserve_kbytes defaults to min(3% of the current process size, 128MB). 792 793 If this is reduced to zero, then the user will be allowed to allocate 794 all free memory with a single process, minus admin_reserve_kbytes. 795 Any subsequent attempts to execute a command will result in 796 "fork: Cannot allocate memory". 797 798 Changing this takes effect whenever an application requests memory. 799 800 ============================================================== 801 802 vfs_cache_pressure 803 ------------------ 804 805 This percentage value controls the tendency of the kernel to reclaim 806 the memory which is used for caching of directory and inode objects. 807 808 At the default value of vfs_cache_pressure=100 the kernel will attempt to 809 reclaim dentries and inodes at a "fair" rate with respect to pagecache and 810 swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer 811 to retain dentry and inode caches. When vfs_cache_pressure=0, the kernel will 812 never reclaim dentries and inodes due to memory pressure and this can easily 813 lead to out-of-memory conditions. Increasing vfs_cache_pressure beyond 100 814 causes the kernel to prefer to reclaim dentries and inodes. 815 816 Increasing vfs_cache_pressure significantly beyond 100 may have negative 817 performance impact. Reclaim code needs to take various locks to find freeable 818 directory and inode objects. With vfs_cache_pressure=1000, it will look for 819 ten times more freeable objects than there are. 820 821 ============================================================= 822 823 watermark_scale_factor: 824 825 This factor controls the aggressiveness of kswapd. It defines the 826 amount of memory left in a node/system before kswapd is woken up and 827 how much memory needs to be free before kswapd goes back to sleep. 828 829 The unit is in fractions of 10,000. The default value of 10 means the 830 distances between watermarks are 0.1% of the available memory in the 831 node/system. The maximum value is 1000, or 10% of memory. 832 833 A high rate of threads entering direct reclaim (allocstall) or kswapd 834 going to sleep prematurely (kswapd_low_wmark_hit_quickly) can indicate 835 that the number of free pages kswapd maintains for latency reasons is 836 too small for the allocation bursts occurring in the system. This knob 837 can then be used to tune kswapd aggressiveness accordingly. 838 839 ============================================================== 840 841 zone_reclaim_mode: 842 843 Zone_reclaim_mode allows someone to set more or less aggressive approaches to 844 reclaim memory when a zone runs out of memory. If it is set to zero then no 845 zone reclaim occurs. Allocations will be satisfied from other zones / nodes 846 in the system. 847 848 This is value ORed together of 849 850 1 = Zone reclaim on 851 2 = Zone reclaim writes dirty pages out 852 4 = Zone reclaim swaps pages 853 854 zone_reclaim_mode is disabled by default. For file servers or workloads 855 that benefit from having their data cached, zone_reclaim_mode should be 856 left disabled as the caching effect is likely to be more important than 857 data locality. 858 859 zone_reclaim may be enabled if it's known that the workload is partitioned 860 such that each partition fits within a NUMA node and that accessing remote 861 memory would cause a measurable performance reduction. The page allocator 862 will then reclaim easily reusable pages (those page cache pages that are 863 currently not used) before allocating off node pages. 864 865 Allowing zone reclaim to write out pages stops processes that are 866 writing large amounts of data from dirtying pages on other nodes. Zone 867 reclaim will write out dirty pages if a zone fills up and so effectively 868 throttle the process. This may decrease the performance of a single process 869 since it cannot use all of system memory to buffer the outgoing writes 870 anymore but it preserve the memory on other nodes so that the performance 871 of other processes running on other nodes will not be affected. 872 873 Allowing regular swap effectively restricts allocations to the local 874 node unless explicitly overridden by memory policies or cpuset 875 configurations. 876 877 ============ End of Document =================================