《MYSQL數(shù)據(jù)庫(kù)MySQL OOM 系統(tǒng)二 OOM Killer》要點(diǎn):
本文介紹了MYSQL數(shù)據(jù)庫(kù)MySQL OOM 系統(tǒng)二 OOM Killer,希望對(duì)您有用。如果有疑問(wèn),可以聯(lián)系我們。
MYSQL應(yīng)用這里就涉及到一個(gè)問(wèn)題,到底Kill掉誰(shuí)呢?一般稍微了解一些Linux內(nèi)核的同學(xué)第一反應(yīng)是誰(shuí)用的最多,就Kill掉誰(shuí).這當(dāng)然是Linux內(nèi)核首先考慮的一種重要因素,但是也不完全是這樣的,我們查一些Linux的內(nèi)核方面的材料,可以知道其實(shí)Kill誰(shuí)是由/proc/<pid>/oom_score來(lái)決定的,這個(gè)值每個(gè)進(jìn)程一個(gè),是由Linux內(nèi)核的oom_badness()函數(shù)負(fù)責(zé)計(jì)算的.那下面我們來(lái)仔細(xì)讀一讀badness()函數(shù).??
MYSQL應(yīng)用在badness()函數(shù)的注釋部門(mén),寫(xiě)明了badness()函數(shù)的處理思路:
MYSQL應(yīng)用???????? 1) we lose the minimum amount of work done
???????? 2) we recover a large amount of memory
???????? 3) we don't kill anything innocent of eating tons of memory
???????? 4) we want to kill the minimum amount of processes (one)
???????? 5) we try to kill the process the user expects us to kill, this? algorithm has been meticulously tuned to meet the principle of least surprise ... (be careful when you change it)
MYSQL應(yīng)用總的來(lái)說(shuō)便是Kill掉最小數(shù)量的進(jìn)程來(lái)獲取最大數(shù)量的內(nèi)存,這與我們Kill掉占用內(nèi)存最大的進(jìn)程是吻合的.
MYSQL應(yīng)用??????? /*
???????? * The memory size of the process is the basis for the badness.
???????? */
MYSQL應(yīng)用???????? points = p->mm->total_vm;
MYSQL應(yīng)用分?jǐn)?shù)的起始是進(jìn)程實(shí)際使用的RAM內(nèi)存,注意這里不包含SWAP,即OOM Killer只會(huì)與進(jìn)程實(shí)際的物理內(nèi)存有關(guān),與Swap是沒(méi)有關(guān)系的,并且我們可以看到,進(jìn)程實(shí)際使用的物理內(nèi)存越多,分?jǐn)?shù)就越高,分?jǐn)?shù)越高就越容易被犧牲掉.
MYSQL應(yīng)用??????? /*
???????? * Processes which fork a lot of child processes are likely
???????? * a good choice. We add the vmsize of the childs if they
???????? * have an own mm. This prevents forking servers to flood the
???????? * machine with an endless amount of childs
???????? */
????????? ...
????????????????? if (chld->mm != p->mm && chld->mm)
??????????????????????? points += chld->mm->total_vm;
MYSQL應(yīng)用這段表現(xiàn)子進(jìn)程占用的內(nèi)存都會(huì)計(jì)算到父進(jìn)程上.
MYSQL應(yīng)用??????? s = int_sqrt(cpu_time);
??????? if (s)
??????????????? points /= s;
??????? s = int_sqrt(int_sqrt(run_time));
??????? if (s)
??????????????? points /= s;
MYSQL應(yīng)用?這注解進(jìn)程占用的CPU時(shí)間越長(zhǎng)或者進(jìn)程運(yùn)行的時(shí)間越長(zhǎng),分?jǐn)?shù)越低,越不容易被Kill掉.
MYSQL應(yīng)用?????? /*
??????? * Niced processes are most likely less important, so double
??????? * their badness points.
??????? */
??????? if (task_nice(p) > 0)
??????????????? points *= 2;
MYSQL應(yīng)用????????? 如果進(jìn)程優(yōu)先級(jí)低(nice值,正值低優(yōu)先級(jí),負(fù)值高優(yōu)先級(jí)),則Point翻倍.
MYSQL應(yīng)用?????? /*
??????? * Superuser processes are usually more important, so we make it
??????? * less likely that we kill those.
??????? */
??????? if (cap_t(p->cap_effective) & CAP_TO_MASK(CAP_SYS_ADMIN) ||
??????????????????????????????? p->uid == 0 || p->euid == 0)
??????????????? points /= 4;
MYSQL應(yīng)用????????? super用戶(hù)的進(jìn)程優(yōu)先級(jí)較低.
MYSQL應(yīng)用??????? /*
???????? * We don't want to kill a process with direct hardware access.
???????? * Not only could that mess up the hardware, but usually users
???????? * tend to only have this flag set on applications they think
???????? * of as important.
???????? */
??????? if (cap_t(p->cap_effective) & CAP_TO_MASK(CAP_SYS_RAWIO))
??????????????? points /= 4;
MYSQL應(yīng)用????????? 直接可以拜訪原始設(shè)備的進(jìn)程優(yōu)先級(jí)較高.
MYSQL應(yīng)用??????? /*
???????? * Adjust the score by oomkilladj.
???????? */
??????? if (p->oomkilladj) {
??????????????? if (p->oomkilladj > 0)
??????????????????????? points <<= p->oomkilladj;
??????????????? else
??????????????????????? points >>= -(p->oomkilladj);
MYSQL應(yīng)用??????? }
MYSQL應(yīng)用每個(gè)進(jìn)程有個(gè)oomkilladj 可以設(shè)置該進(jìn)程被kill的優(yōu)先級(jí),這個(gè)參數(shù)看上去對(duì)Point影響還是比擬大的,oomkilladj 最大+15,最小是-17,越大越容易被干掉,這個(gè)值由于是移位運(yùn)算,所以影響還是比擬大的.
MYSQL應(yīng)用下面我寫(xiě)個(gè)小程序?qū)嶒?yàn)一下:
MYSQL應(yīng)用
#define MEGABYTE 1024*1024*1024
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
int main(int argc, char *argv[])
{
void *myblock = NULL;
myblock = (void *) malloc(MEGABYTE);
printf("Currently allocating 1GB\n");
sleep(1);
int count = 0;
while( count < 10)
{
memset(myblock,1,100*1024*1024);
myblock = myblock + 100*1024*1024;
count++;
printf("Currently allocating %d00 MB\n",count);
sleep(10);
}
exit(0);
}
MYSQL應(yīng)用上面的程序先申請(qǐng)一個(gè)1G的內(nèi)存空間,然后100M為單位,填充這些內(nèi)存空間.在一個(gè)2G內(nèi)存,400M Swap空間的機(jī)器上跑3個(gè)上面的進(jìn)程.我們看一下運(yùn)行結(jié)果:
MYSQL應(yīng)用? ?
MYSQL應(yīng)用? ? ?test1、test2、test3分別申請(qǐng)了1G的虛擬內(nèi)存空間(VIRT),然后每隔10s,實(shí)際占用的RAM空間就增長(zhǎng)100M(RES).
MYSQL應(yīng)用? ??
MYSQL應(yīng)用? ??當(dāng)物理內(nèi)存空間不敷時(shí),OS開(kāi)始進(jìn)行Swap,可用的Swap空間開(kāi)始減少.
MYSQL應(yīng)用? ?
? ??
MYSQL應(yīng)用? ? ?當(dāng)內(nèi)存是在沒(méi)有可分配的空間時(shí),test1進(jìn)程被操作系統(tǒng)Kill掉了.dmesg 我們可以看到,test1進(jìn)程被OS Kill掉,同時(shí)oom_score為1000.
MYSQL應(yīng)用?????這3個(gè)進(jìn)程的oom_adj全部都是默認(rèn)值0.下面我們來(lái)實(shí)驗(yàn)一下設(shè)置了oom_adj的效果.重新啟動(dòng)3個(gè)進(jìn)程,然后我們看到test2的PID是12640
MYSQL應(yīng)用? ?? ??
MYSQL應(yīng)用? ?我們運(yùn)行一下下面的語(yǔ)句
MYSQL應(yīng)用???echo 15 > /proc/12640/oom_adj
MYSQL應(yīng)用? 一段時(shí)間后,我們看到Swap空間急劇減少,基本上OS OOM_Killer要開(kāi)動(dòng)了.?
MYSQL應(yīng)用? ?
MYSQL應(yīng)用? ? ?果然,不出意料,12640進(jìn)程被kill掉了.
MYSQL應(yīng)用???? 所以為了避免本身需要的進(jìn)程被kill掉,可以通過(guò)設(shè)置進(jìn)程的oom_adj來(lái)實(shí)現(xiàn).當(dāng)然,有的人會(huì)說(shuō),這一切都是超售引起的,既然Linux提供了overcommit_memory可以禁用overcommit特性,那為什么不禁用呢.這有利也有弊,一旦禁用overcommit,就意味著MySQL根本無(wú)法申請(qǐng)超過(guò)實(shí)際內(nèi)存的空間,而在MySQL中,存在很多動(dòng)態(tài)申請(qǐng)內(nèi)存空間的地方,如果申請(qǐng)不到,MySQL就會(huì)Crash,這大大增加了MySQL宕機(jī)的風(fēng)險(xiǎn),這也是Linux為什么要overcommit的原因.
MYSQL應(yīng)用????? 有了上面的闡發(fā),我們不難看出,如果在不設(shè)置oom_adj的前提下,MySQL一般都會(huì)成為OOM_Killer的首選對(duì)象,因?yàn)镸ySQL一般都是內(nèi)存的最大占用者.那作為MySQL,我們?nèi)绾伪M量的去規(guī)避被Kill的風(fēng)險(xiǎn)呢,下一章我們將重點(diǎn)從MySQL的角度闡發(fā)如何規(guī)避OOM.
歡迎參與《MYSQL數(shù)據(jù)庫(kù)MySQL OOM 系統(tǒng)二 OOM Killer》討論,分享您的想法,維易PHP學(xué)院為您提供專(zhuān)業(yè)教程。
轉(zhuǎn)載請(qǐng)注明本頁(yè)網(wǎng)址:
http://www.snjht.com/jiaocheng/13583.html