mongodb - 执行 mongodump 后 Mongod 服务崩溃

Question

我执行了此查询

查询:mongodump --db=elastic --collection=q_moonx_notifications_2019-08-26 --out=/home/centos/mongo-dump/

在我的 mongo 数据库上转储一个周围的集合

集合名称:q_moonx_notifications_2019-08-26
大小:80GB存储大小:23.6GB

执行此操作后不久，mongod 服务崩溃了。我通过/var/log/messages 找到了问题。我知道这是由于“内存不足”问题而发生的。有人可以帮我看看这是怎么发生的，以及如何在不影响我正在运行的 mongo 服务的情况下转储单个集合。

机器有 32 GB 内存和 0 swp。

/var/log/messages 内容

Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb61e41>] dump_stack+0x19/0x1b
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb5c86a>] dump_header+0x90/0x229
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd700bcb>] ? cred_has_capability+0x6b/0x120
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5ba4e4>] oom_kill_process+0x254/0x3d0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd700c9c>] ? selinux_capable+0x1c/0x40
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5bad26>] out_of_memory+0x4b6/0x4f0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb5d36e>] __alloc_pages_slowpath+0x5d6/0x724
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5c1105>] __alloc_pages_nodemask+0x405/0x420
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd60df68>] alloc_pages_current+0x98/0x110
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5b6347>] __page_cache_alloc+0x97/0xb0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5b8fa8>] filemap_fault+0x298/0x490
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffc0400d0e>] __xfs_filemap_fault+0x7e/0x1d0 [xfs]
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffc0400f0c>] xfs_filemap_fault+0x2c/0x30 [xfs]
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5e444a>] __do_fault.isra.59+0x8a/0x100
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5e49fc>] do_read_fault.isra.61+0x4c/0x1b0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5e93a4>] handle_pte_fault+0x2f4/0xd10
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd50cbf8>] ? get_futex_key+0x1c8/0x2c0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbd5ebedd>] handle_mm_fault+0x39d/0x9b0
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb6f5e3>] __do_page_fault+0x203/0x500
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb6f915>] do_page_fault+0x35/0x90
Oct 11 07:07:33 ip-1.23.345.678 kernel: [<ffffffffbdb6b758>] page_fault+0x28/0x30
Oct 11 07:07:33 ip-1.23.345.678 kernel: Mem-Info:
Oct 11 07:07:33 ip-1.23.345.678 kernel: active_anon:7972560 inactive_anon:30565 isolated_anon:0#012 active_file:2831 inactive_file:4651 isolated_file:0#012 unevictable:0 dirty:0 writeback:5 unstable:0#012 slab_reclaimable:42065 slab_unreclaimable:12016#012 mapped:18928 shmem:55424 pagetables:19349 bounce:0#012 free:49154 free_pcp:558 free_cma:0
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 DMA free:15904kB min:32kB low:40kB high:48kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15988kB managed:15904kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? yes
Oct 11 07:07:33 ip-1.23.345.678 kernel: lowmem_reserve[]: 0 3597 31992 31992
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 DMA32 free:121020kB min:7596kB low:9492kB high:11392kB active_anon:3397760kB inactive_anon:11020kB active_file:284kB inactive_file:508kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:3915776kB managed:3684320kB mlocked:0kB dirty:0kB writeback:0kB mapped:5892kB shmem:14892kB slab_reclaimable:131668kB slab_unreclaimable:5368kB kernel_stack:1472kB pagetables:7364kB unstable:0kB bounce:0kB free_pcp:1184kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:2829 all_unreclaimable? yes
Oct 11 07:07:33 ip-1.23.345.678 kernel: lowmem_reserve[]: 0 0 28394 28394
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 Normal free:66532kB min:59952kB low:74940kB high:89928kB active_anon:28492480kB inactive_anon:111240kB active_file:11040kB inactive_file:11684kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:29622272kB managed:29079004kB mlocked:0kB dirty:0kB writeback:20kB mapped:69820kB shmem:206804kB slab_reclaimable:36592kB slab_unreclaimable:42696kB kernel_stack:6720kB pagetables:70032kB unstable:0kB bounce:0kB free_pcp:3320kB local_pcp:0kB free_cma:0kB writeback_tmp:0kB pages_scanned:8574 all_unreclaimable? no
Oct 11 07:07:33 ip-1.23.345.678 kernel: lowmem_reserve[]: 0 0 0 0
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 DMA: 0*4kB 0*8kB 0*16kB 1*32kB (U) 2*64kB (U) 1*128kB (U) 1*256kB (U) 0*512kB 1*1024kB (U) 1*2048kB (M) 3*4096kB (M) = 15904kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 DMA32: 2724*4kB (UEM) 1319*8kB (UEM) 2734*16kB (UEM) 1065*32kB (UEM) 296*64kB (UEM) 32*128kB (UEM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 122312kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 Normal: 10268*4kB (UEM) 3615*8kB (UEM) 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 69992kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=1048576kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: 61902 total pagecache pages
Oct 11 07:07:33 ip-1.23.345.678 kernel: 0 pages in swap cache
Oct 11 07:07:33 ip-1.23.345.678 kernel: Swap cache stats: add 0, delete 0, find 0/0
Oct 11 07:07:33 ip-1.23.345.678 kernel: Free swap = 0kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: Total swap = 0kB
Oct 11 07:07:33 ip-1.23.345.678 kernel: 8388509 pages RAM
Oct 11 07:07:33 ip-1.23.345.678 kernel: 0 pages HighMem/MovableOnly
Oct 11 07:07:33 ip-1.23.345.678 kernel: 193702 pages reserved
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ pid ] uid tgid total_vm rss nr_ptes swapents oom_score_adj name
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 2415] 0 2415 55590 32605 114 0 0 systemd-journal
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 2456] 0 2456 11953 611 25 0 -1000 systemd-udevd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 2704] 0 2704 15511 170 29 0 -1000 auditd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4346] 32 4346 18412 189 38 0 0 rpcbind
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4464] 81 4464 16600 204 34 0 -900 dbus-daemon
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4532] 998 4532 29446 143 29 0 0 chronyd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4588] 0 4588 6652 156 19 0 0 systemd-logind
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4589] 999 4589 153057 1381 63 0 0 polkitd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4592] 0 4592 5416 101 14 0 0 irqbalance
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4596] 0 4596 50404 162 38 0 0 gssproxy
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 4940] 0 4940 26839 508 51 0 0 dhclient
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5035] 0 5035 143455 3309 99 0 0 tuned
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5102] 0 5102 31253 535 58 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5103] 995 5103 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5104] 995 5104 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5105] 995 5105 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5106] 995 5106 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5107] 995 5107 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5108] 995 5108 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5109] 995 5109 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5110] 995 5110 31375 641 59 0 0 nginx
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5183] 0 5183 22603 310 42 0 0 master
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5189] 89 5189 22673 286 43 0 0 qmgr
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5231] 997 5231 4365796 4120011 8150 0 0 mongod
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5295] 0 5295 104225 16973 121 0 0 rsyslogd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5297] 0 5297 28189 267 58 0 -1000 sshd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5337] 0 5337 31580 194 18 0 0 crond
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5343] 0 5343 27523 50 10 0 0 agetty
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5347] 0 5347 27523 50 13 0 0 agetty
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 5662] 27 5662 691356 97174 303 0 0 mysqld
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 7361] 996 7361 12739660 2366239 5813 0 0 java
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 7462] 996 7462 17192 173 30 0 0 controller
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 7661] 1000 7661 3588101 1239871 2579 0 0 java
Oct 11 07:07:33 ip-1.23.345.678 kernel: [14974] 0 14974 371181 6509 107 0 0 metricbeat
Oct 11 07:07:33 ip-1.23.345.678 kernel: [ 6781] 994 6781 430375 60775 604 0 0 node
Oct 11 07:07:33 ip-1.23.345.678 kernel: [24008] 89 24008 22629 301 47 0 0 pickup
Oct 11 07:07:33 ip-1.23.345.678 kernel: [25163] 0 25163 39154 367 77 0 0 sshd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [25167] 1000 25167 39154 366 74 0 0 sshd
Oct 11 07:07:33 ip-1.23.345.678 kernel: [25168] 1000 25168 28893 149 15 0 0 bash
Oct 11 07:07:33 ip-1.23.345.678 kernel: [28554] 1000 28554 260690 34619 132 0 0 mongodump
Oct 11 07:07:33 ip-1.23.345.678 kernel: Out of memory: Kill process 5231 (mongod) score 503 or sacrifice child
Oct 11 07:07:33 ip-1.23.345.678 kernel: Killed process 5231 (mongod) total-vm:17463184kB, anon-rss:16480044kB, file-rss:0kB, shmem-rss:0kB
Oct 11 07:07:34 ip-1.23.345.678 systemd: mongod.service: main process exited, code=killed, status=9/KILL
Oct 11 07:07:34 ip-1.23.345.678 systemd: Unit mongod.service entered failed state.
Oct 11 07:07:34 ip-1.23.345.678 systemd: mongod.service failed.
Oct 11 07:07:39 ip-1.23.345.678 systemd-logind: New session 1220 of user centos.


Answer 1

您遇到了一个称为“OOM-Killer”的“问题”。

引用自 MongoDB 文档中恰当命名为 Production notes 的部分:

Swap

Assign swap space for your systems. Allocating swap space can avoid issues with memory contention and can prevent the OOM Killer on Linux systems from killing mongod.

For the WiredTiger storage engine, given sufficient memory pressure, WiredTiger may store data in swap space.

(我强调的)

基本上发生的事情是系统的内存压力变得严重，内核决定终止一个进程以释放一些 RAM，以确保系统仍然可以运行。它通过确定任务的“坏程度”来做到这一点

badness_for_task = total_vm_for_task / (sqrt(cpu_time_in_seconds) *
sqrt(sqrt(cpu_time_in_minutes)))

并杀死最糟糕的任务。请参阅https://www.kernel.org/doc/gorman/html/understand/understand016.html了解详情。

要点:MongoDB 消耗的 RAM 可能比服务器上的任何其他进程多几个数量级，因此当内核无法选择交换某些数据时，它会被 OOM Killer 杀死，从而确保基本系统任务仍然可以执行运行。

这种行为基本上可以通过分配交换来防止，这就是为什么它会相应地记录在生产说明中。