线上 OutOfMemoryError 了，没头绪... ...

V2EX = way to explore

V2EX 是一个关于分享和探索的地方

已注册用户请登录

这是一个创建于 1995 天前的主题，其中的信息可能已经有所发展或是发生改变。

前提

查看了应用日志，是凌晨就已经 OOM 了，但是直到早上八点多客服反馈登录不了才发现问题，当时在路上，于是让运维 dump 了堆栈和线程信息，重启了服务器。

jvm 配置：

-Xmx2000M -Xms2000M -Xmn500M -XX:PermSize=250M -XX:MaxPermSize=250M -Xss256K -XX:-UseGCOverheadLimit -Djava.net.preferIPv4Stack=true -Djava.net.preferIPv4Addresses -XX:+DisableExplicitGC -XX:SurvivorRatio=1 -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:+CMSParallelRemarkEnabled -XX:+UseCMSCompactAtFullCollection -XX:CMSFullGCsBeforeCompaction=0 -XX:+CMSClassUnloadingEnabled -XX:LargePageSizeInBytes=128M -XX:+UseFastAccessorMethods -XX:+UseCMSInitiatingOccupancyOnly -XX:CMSInitiatingOccupancyFraction=60 -XX:SoftRefLRUPolicyMSPerMB=0 -XX:+PrintClassHistogram -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintHeapAtGC -Xloggc:/data/apps/apache-tomcat-6.0.36/logs/gc.log

jvm.heap

2020-05-07 08:46: using parallel threads in the new generation. using thread-local object allocation. Concurrent Mark-Sweep GC Heap Configuration: MinHeapFreeRatio = 40 MaxHeapFreeRatio = 70 MaxHeapSize = 2097152000 (2000.0MB) NewSize = 524288000 (500.0MB) MaxNewSize = 524288000 (500.0MB) OldSize = 5439488 (5.1875MB) NewRatio = 2 SurvivorRatio = 1 PermSize = 262144000 (250.0MB) MaxPermSize = 262144000 (250.0MB) Heap Usage: New Generation (Eden + 1 Survivor Space): capacity = 349569024 (333.375MB) used = 137905504 (131.51693725585938MB) free = 211663520 (201.85806274414062MB) 39.45014990801931% used Eden Space: capacity = 174850048 (166.75MB) used = 129558040 (123.5561752319336MB) free = 45292008 (43.193824768066406MB) 74.09665681075478% used From Space: capacity = 174718976 (166.625MB) used = 8347464 (7.960762023925781MB) free = 166371512 (158.66423797607422MB) 4.777651627262284% used To Space: capacity = 174718976 (166.625MB) used = 0 (0.0MB) free = 174718976 (166.625MB) 0.0% used concurrent mark-sweep generation: capacity = 1572864000 (1500.0MB) used = 532712344 (508.0340805053711MB) free = 1040151656 (991.9659194946289MB) 33.86893870035807% used Perm Generation: capacity = 262144000 (250.0MB) used = 149215056 (142.3025665283203MB) free = 112928944 (107.69743347167969MB) 56.921026611328124% used 2020-05-07 08:46: using thread-local object allocation. Parallel GC with 4 thread(s) Heap Configuration: MinHeapFreeRatio = 40 MaxHeapFreeRatio = 70 MaxHeapSize = 1566572544 (1494.0MB) NewSize = 1310720 (1.25MB) MaxNewSize = 17592186044415 MB OldSize = 5439488 (5.1875MB) NewRatio = 2 SurvivorRatio = 8 PermSize = 21757952 (20.75MB) MaxPermSize = 174063616 (166.0MB) Heap Usage: PS Young Generation Eden Space: capacity = 1966080 (1.875MB) used = 628456 (0.5993423461914062MB) free = 1337624 (1.2756576538085938MB) 31.964925130208332% used From Space: capacity = 196608 (0.1875MB) used = 98304 (0.09375MB) free = 98304 (0.09375MB) 50.0% used To Space: capacity = 196608 (0.1875MB) used = 0 (0.0MB) free = 196608 (0.1875MB) 0.0% used PS Old Generation capacity = 65273856 (62.25MB) used = 4938552 (4.709770202636719MB) free = 60335304 (57.54022979736328MB) 7.565895907850151% used PS Perm Generation capacity = 25296896 (24.125MB) used = 25255776 (24.085784912109375MB) free = 41120 (0.039215087890625MB) 99.8374504128886% used

应用日志

jstack 信息

2020-05-07 08:47 cpu_busy: 1 "scheduler-9" prio=10 tid=0x00007f962c6e6000 nid=0x3a0d waiting on condition [0x00007f962437f000] java.lang.Thread.State: TIMED_WAITING (parking) at sun.misc.Unsafe.park(Native Method) - parking to wait for <0x0000000796111028> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject) at java.util.concurrent.locks.LockSupport.parkNanos(LockSupport.java:226) at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2081) at java.util.concurrent.DelayQueue.take(DelayQueue.java:193) at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:688) at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:681) at java.util.concurrent.ThreadPoolExecutor.getTask(ThreadPoolExecutor.java:1069) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1131) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615) at java.lang.Thread.run(Thread.java:679) "scheduler-8" prio=10 tid=0x00007f962cb95800 nid=0x3a0c waiting on condition [0x00007f96248d4000] java.lang.Thread.State: TIMED_WAITING (parking) at sun.misc.Unsafe.park(Native Method) - parking to wait for <0x0000000796111028> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject) at java.util.concurrent.locks.LockSupport.parkNanos(LockSupport.java:226) at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2081) at java.util.concurrent.DelayQueue.take(DelayQueue.java:193) at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:688) at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:681) at java.util.concurrent.ThreadPoolExecutor.getTask(ThreadPoolExecutor.java:1069) at java.util.concurrent.ThreadPoolExecutor.runWorkr(ThreadPoolExecutor.java:1131) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615) at java.lang.Thread.run(Thread.java:679) "scheduler-7" prio=10 tid=0x00007f962c7c6000 nid=0x3a0b waiting on condition [0x00007f962468b000] java.lang.Thread.State: TIMED_WAITING (parking) at sun.misc.Unsafe.park(Native Method) 2020-05-07 08:47 cpu_busy: 1 "VM Thread" prio=10 tid=0x00007f962c0fe800 nid=0x38e1 runnable "Gang worker#0 (Parallel GC Threads)" prio=10 tid=0x00007f962c010800 nid=0x38dc runnable "Gang worker#1 (Parallel GC Threads)" prio=10 tid=0x00007f962c012800 nid=0x38dd runnable "Gang worker#2 (Parallel GC Threads)" prio=10 tid=0x00007f962c014800 nid=0x38de runnable "Gang worker#3 (Parallel GC Threads)" prio=10 tid=0x00007f962c016000 nid=0x38df runnable "Concurrent Mark-Sweep GC Thread" prio=10 tid=0x00007f962c0a5000 nid=0x38e0 runnable "VM Periodic Task Thread" prio=10 tid=0x00007f962c258000 nid=0x39b4 waiting on condition JNI global references: 1952 2020-05-07 08:47 cpu_busy: 1 "DefaultQuartzScheduler_QuartzSchedulerThread" prio=10 tid=0x00007f1c8c39c000 nid=0x6687 waiting on condition [0x00007f1c4b9f8000] java.lang.Thread.State: TIMED_WAITING (sleeping) at java.lang.Thread.sleep(Native Method) at org.quartz.core.QuartzSchedulerThread.run(QuartzSchedulerThread.java:275) "DefaultQuartzScheduler_Worker-10" prio=10 tid=0x00007f1c8c3a8000 nid=0x6686 in Object.wait() [0x00007f1c4baf9000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a06000> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-9" prio=10 tid=0x00007f1c8c3d1000 nid=0x6685 in Object.wait() [0x00007f1c4bbfa000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a001a8> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-8" prio=10 tid=0x00007f1c8c1e7800 nid=0x6684 in Object.wait() [0x00007f1c4bcfb000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a04020> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-7" prio=10 tid=0x00007f1c8c3a5000 nid=0x6683 in Object.wait() [0x00007f1c4bdfc000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a00228> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-6" prio=10 tid=0x00007f1c8c241000 nid=0x6682 in Object.wait() [0x00007f1c3bffe000] java.lang.Thread.State: TIMED_WAITING (on object monitor) 2020-05-07 08:47 cpu_busy: 1 "Attach Listener" daemon prio=10 tid=0x00007f1c68001000 nid=0x30d waiting on condition [0x0000000000000000] java.lang.Thread.State: RUNNABLE "DestroyJavaVM" prio=10 tid=0x00007f1c8c008000 nid=0x665b waiting on condition [0x0000000000000000] java.lang.Thread.State: RUNNABLE "DefaultQuartzScheduler_QuartzSchedulerThread" prio=10 tid=0x00007f1c8c39c000 nid=0x6687 waiting on condition [0x00007f1c4b9f8000] java.lang.Thread.State: TIMED_WAITING (sleeping) at java.lang.Thread.sleep(Native Method) at org.quartz.core.QuartzSchedulerThread.run(QuartzSchedulerThread.java:275) "DefaultQuartzScheduler_Worker-10" prio=10 tid=0x00007f1c8c3a8000 nid=0x6686 in Object.wait() [0x00007f1c4baf9000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a06000> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-9" prio=10 tid=0x00007f1c8c3d1000 nid=0x6685 in Object.wait() [0x00007f1c4bbfa000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a001a8> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-8" prio=10 tid=0x00007f1c8c1e7800 nid=0x6684 in Object.wait() [0x00007f1c4bcfb000] java.lang.Thread.State: TIMED_WAITING (on object monitor) at java.lang.Object.wait(Native Method) at org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:523) - locked <0x00000000a2a04020> (a org.quartz.simpl.SimpleThreadPool$WorkerThread) "DefaultQuartzScheduler_Worker-7" prio=10 tid=0x00007f1c8c3a5000 nid=0x6683 in Object.wait() [0x00007f1c4bdfc000] java.lang.Thread.State: TIMED_WAITING (on object monitor)

用 jhat -J-Xmx50m jvm.dump 查看所有类的实例数

这是我第一次碰到 OOM，我不理解的是为什么我看 jvm.heap 里面，堆内存使用率不是百分百，但请求依然是 OOM 呢？这些堆栈文件是运维用脚本生成的，必要的话我可以把脚本贴上来。

我会继续排查问题，希望哪位看到了也能帮我瞅瞅，非常感谢。

第 1 条附言 2020-05-07 14:21:01 +08:00

ps:

发现日志的打印时间顺序有问题，不知道这个会不会可以找到什么线索？

第 2 条附言 2020-05-07 20:42:30 +08:00

pps

今晚准备做好再次OOM的准备，做了以下工作

给JVM增加了参数： -XX:+HeapDumpOnOutOfMemoryError，然后一会儿十点半重启下服务器，下次发生OOM及时拿到最完整的dump信息（有可能重启一次会清空内存让oom推迟发生，但也还是得加）
本地用jconsole连上线上服务器，实时监控内存各个区的状态。
已经给运维同事打过招呼，如果发生故障可能要打电话给他重启服务。当然故障时应该没人在用服务，我会稍微研究会儿再重启

第 3 条附言 2020-05-07 20:45:43 +08:00

哪位大佬如果有好的建议，赶紧提一下赶在故障前我准备下。

第 4 条附言 2020-05-08 10:39:22 +08:00

大概知道出现问题的原因了，今天先修复代码，后续会总结个文章贴出来。

47 条回复 2020-05-08 12:01:55 +08:00

alexfarm

2020-05-07 12:46:47 +08:00

以前用过个工具 eclipse memory analyzer 去看 dump 的内容，稍微直观详细点，主要看看堆内存占用高的类

Lax

2020-05-07 12:47:57 +08:00

不太懂 java，不过看你输出里有个 PS Perm Generation 已经满了。

aircjm

2020-05-07 12:48:17 +08:00

这个应该是对象太大导致的吧?? 你可以用 eclipse 的那个分析工具分析下你的 dump 文件找下

Solace202

2020-05-07 12:58:47 +08:00

@alexfarm
@aircjm 谢谢，刚下载了这个工具，学着先用下。

Finest

2020-05-07 13:04:54 +08:00

你这里是两个进程？ jmap 输出的第二段，明显 jvm 的内存配置都跟你贴的不一样，Perm Gen 才 24m

aircjm

2020-05-07 13:07:22 +08:00

@Solace202 客气，网上有教程的，可以先试试

lovedebug

2020-05-07 13:17:17 +08:00

我跟楼主说一个自己碰到过的。线上爆发的及其容易忽略的导致 oom 的点，所有 debug 和 info 级别 log 中不要打印整个 List/Set/Map，这些在运行中实际也是会计算的。

lovedebug &nsp;

2020-05-07 13:18:10 +08:00

另外楼主应该为这个服务设置 health check api，可以今早发现问题。

lovedebug

2020-05-07 13:19:35 +08:00

PPS：先升级机器内存起一台新机器线上泡泡看看，这台出问题的机器可以从集群中移除仔细分析

pmispig

2020-05-07 13:34:49 +08:00

-XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/logs/gc.hprof
加上这个

Livid

MOD

PRO

2020-05-07 13:35:28 +08:00

你们有在用 Sentry.io 么？

Solace202

2020-05-07 13:38:52 +08:00

@hand515 这是使用了两个 GC （ Mark-Sweep GC 和 Parallel GC ），jvm 里面配置是给第一个 GC 配的
![]( https://i.bmp.ovh/imgs/2020/05/56930920c1ac5a20.png)

yamasa

2020-05-07 13:39:41 +08:00

jmap -histo <pid>会按 mem size 对堆内对象排序，先看看前几名分别是啥。但是这种方式看不出对象树，所以比较大可能还是需要用 jmap dump 一个 bin 或者 hprof 下来，用 eclipse mem analyzer 直接分析。这个 analyzer 可以对一个 dump 出具 mem leak 的报表，要仔细看看，找到了高内存占用的类看看对象树，是不是你们哪里的业务代码造成了内存泄漏。

Solace202

2020-05-07 13:40:00 +08:00

@hand515 贴个配置和 jmap 的对应![关系图]( https://i.bmp.ovh/imgs/2020/05/56930920c1ac5a20.png)

Solace202

2020-05-07 13:41:43 +08:00

@Livid 没用过

Solace202

2020-05-07 13:42:22 +08:00

@pmispig 嗯，这个之前没加，我让运维加上，但是得再出个问题才能拿到。。。

yamasa

2020-05-07 13:42:59 +08:00

话说你们的 GC 是 CMS,可以用这个工具看看 https://github.com/DarLiner/vjtools/tree/master/vjmap 。如果是 G1 的话反而用不了。这是可以分代的 jmap，如果 jmap 给出的分布看不出来问题就上这个，一个一个代区来分析。

Solace202

2020-05-07 13:54:33 +08:00

@pmispig 已经让运维加了，我们有 GC 的日志，但是没有日期，这个能否定位问题呢？可以的话我让在 JVM 把 gc 的日期和时间也加上？

limbo0

2020-05-07 14:11:53 +08:00

除了堆内存, 也得考虑下堆外内存,是不是有大量的网络请求

coolcfan

2020-05-07 14:13:14 +08:00 via Android

如果是 PermGen 满，结合日志内容，可以关注下应用里的反射操作或者是否存在动态的类生成……

Solace202

2020-05-07 14:15:18 +08:00

@limbo0 凌晨一般产品就没人使用了，只是凌晨一般会议很多定时任务开始执行

25OHd2qObJmJ6P10

2020-05-07 14:17:38 +08:00

从应用日志来看是 spring 注入一个类的时候报了堆空间不足，spring 注入属性会去加载相应的类到永久区，根据你的日志分析，你的 jdk 应该是 1.8 以下，而且你的 jvm 内存空间也显示你的永久区只有 24M,使用达到 99%,所以初步分析就可以增大永久区 -XX:PermSize=75M -XX:MaxPermSize=100M 尝试，然后再根据 jstat 和 jmap 分析内存变化，确定原因以及调整最合适的 jvm 参数

Solace202

2020-05-07 14:26:56 +08:00

@rayduan 我是有个地方不理解，两个 GC，为什么注入的类会放到下面这个 GC 的 Perm Generation 而不是上面的那个 Perm Generation ？

zwx327634

2020-05-07 14:31:50 +08:00

如果是 jdk1.7,升到 1.8 就没问题了，永久代变成元空间，放在堆里了。快速的话就是像楼上，增加永久代空间

Solace202

2020-05-07 14:48:23 +08:00

@zwx327634 jdk1.6，瑟瑟发抖，老项目快十年了

1ffree

2020-05-07 15:05:50 +08:00

好奇应用运行期间还有 autowire 是啥使用场景呢

pangleon

2020-05-07 15:07:44 +08:00

你这个提示的是 heap space 而不是 permgen space，后者是永久代满了，那么基本上可以定位到你们的程序创建了大量的临时对象

如果真是凌晨 OOM，凌晨也不是用户高峰那么基本可以确定是定时任务导致的。
你可以查询下有哪些定时任务在运行，他们是否创建了大量的 LIST,MAP 这种集合类对象或者拼接了大量字符串对象。

pmispig

2020-05-07 15:15:31 +08:00

@Solace202 -XX:HeapDumpPath=/logs/gc.hprof 看看能不能重现一次，拿到这个 dump 出来分析就比较好定位