Java新生代回收一瞥-垃圾回收用arthas看dashboard
效果
一直刷新列表生成新对象 "人工诱发垃圾回收"
看两次Memory里面的eden_space used 从136M变成1M
Memory used total max usage GC
heap 253M 495M 1979M 12.78% gc.copy.count 61
tenured_gen 110M 341M 1365M 8.12% gc.copy.time(ms) 2096
eden_space 136M 136M 546M 24.91% gc.marksweepcompact.count 5
survivor_space 6M 17M 68M 8.99% gc.marksweepcompact.time(ms) 1234
nonheap 212M 216M -1 98.08%
codeheap_'non-nmethods' 1M 4M 7M 20.05%
metaspace 140M 141M -1 99.33%
codeheap_'profiled_nmethods' 40M 40M 116M 34.55%
compressed_class_space 14M 14M 1024M 1.42%
codeheap_'non-profiled_nmethods' 15M 15M 116M 13.34%
mapped 0K 0K - 0.00%
direct 316M 316M - 100.00%
mapped - 'non-volatile memory' 0K 0K - 0.00%
Runtime
os.name Linux
os.version 6.8.0-90-generic
java.version 24.0.1
java.home /opt/java/openjdk
systemload.average 0.13
processors 32
timestamp/uptime Fri Jan 23 10:37:51 CST 2026/689058s
ID NAME GROUP PRIORITY STATE %CPU DELTA_TIME TIME INTERRUPTED DAEMON
274 reactor-http-epoll-21 main 5 RUNNABLE 0.87 0.008 0:0.313 false true
248 parallel-30 main 5 WAITING 0.7 0.007 0:0.162 false true
78 redisson-netty-7-2 main 5 RUNNABLE 0.46 0.004 0:6.164 false false
24765 Timer-for-arthas-dashboard-17f047bb-e86a-456e-962d- system 5 RUNNABLE 0.44 0.004 0:0.041 false true
24753 arthas-NettyHttpTelnetBootstrap-3-4 system 5 RUNNABLE 0.36 0.003 0:0.051 false true
82 redisson-netty-7-6 main 5 RUNNABLE 0.32 0.003 0:6.123 false false
31 Notification Thread system 9 RUNNABLE 0.16 0.001 0:0.118 false true
85 redisson-netty-7-8 main 5 RUNNABLE 0.12 0.001 0:6.143 false false
51 redisson-timer-6-1 main 5 TIMED_WAITING 0.05 0.000 4:37.761 false false
76 redisson-timer-9-1 main 5 TIMED_WAITING 0.04 0.000 4:34.719 false false
65 reactor-http-epoll-1 main 5 RUNNABLE 0.04 0.000 0:0.601 false true
66 Thread-6 main 5 WAITING 0.03 0.000 0:2.148 false true
38 AsyncAppender-Worker-ASYNC_APP main 5 WAITING 0.02 0.000 0:0.573 false true
19218 ForkJoinPool-1-worker-16 CarrierThreads 5 TIMED_WAITING 0.02 0.000 0:1.138 false true
50 MaintenanceTimer-1-thread-1 main 5 TIMED_WAITING 0.01 0.000 0:1.191 false true
39 AsyncAppender-Worker-ASYNC_ERROR main 5 WAITING 0.01 0.000 0:0.217 false true
11 Reference Handler system 10 RUNNABLE 0.01 0.000 0:0.027 false true
32 Common-Cleaner InnocuousThreadGroup 8 TIMED_WAITING 0.01 0.000 0:0.680 false true
96 workIdHolder-8-1768446835584 main 5 TIMED_WAITING 0.01 0.000 0:27.964 false false
95 Thread-7 main 5 TIMED_WAITING 0.0 0.000 0:19.333 false true
12 Finalizer system 8 WAITING 0.0 0.000 0:0.000 false true
13 Signal Dispatcher system 9 RUNNABLE 0.0 0.000 0:0.000 false true
33 Cleaner-0 InnocuousThreadGroup 8 TIMED_WAITING 0.0 0.000 0:0.711 false true
35 Attach Listener system 9 RUNNABLE 0.0 0.000 0:0.038 false true
43 parallel-1 main 5 WAITING 0.0 0.000 0:0.470 false true
44 parallel-2 main 5 WAITING 0.0 0.000 0:0.445 false true
46 AsyncGetter-2-thread-1 main 5 WAITING 0.0 0.000 0:0.180 false true
Memory used total max usage GC
heap 119M 495M 1979M 6.02% gc.copy.count 62
tenured_gen 110M 341M 1365M 8.12% gc.copy.time(ms) 2131
eden_space 1M 136M 546M 0.22% gc.marksweepcompact.count 5
survivor_space 7M 17M 68M 10.58% gc.marksweepcompact.time(ms) 1234
nonheap 212M 216M -1 98.06%
codeheap_'non-nmethods' 1M 4M 7M 20.05%
metaspace 140M 141M -1 99.34%
codeheap_'profiled_nmethods' 40M 40M 116M 34.57%
compressed_class_space 14M 14M 1024M 1.42%
codeheap_'non-profiled_nmethods' 15M 15M 116M 13.34%
mapped 0K 0K - 0.00%
direct 316M 316M - 100.00%
mapped - 'non-volatile memory' 0K 0K - 0.00%
AI帮我分析
1. 抓捕“消失的瞬间”
刷新前：eden_space 是 136M (24.91%)。注意，这里的 total 也是 136M。这意味着 Eden 区已经顶到了天花板，处于“满载”状态。
刷新后：eden_space 变成了 1M (0.22%)。
发生了什么：当你持续刷新列表时，Spring Boot 会不断创建 DTO、数据库连接对象、JSON 序列化字符数组等。因为 Eden 区已经满了，JVM 此时触发了 Minor GC (gc.copy.count 从 61 变成了 62)。
2. 垃圾回收的“证据链”
回收了多少？ Eden 区从 136M 掉到 1M，清理掉了约 135M 的临时对象。
花了多久？ gc.copy.time 从 2096ms 变为 2131ms，说明这次清理只用了 35 毫秒。
幸存者搬家了吗？ * survivor_space 从 6M 变到了 7M。这说明刷新产生的对象中，有 1M 左右的对象（可能是你的 Session、缓存或其他长生命周期对象）没被回收，而是按照 2.4 垃圾回收算法 里的“复制算法”，被搬到了 Survivor 区。
tenured_gen（老年代）依然是 110M。说明你的列表数据都是“朝生夕死”的，没有直接晋升到老年代。
命令
linux的docker里面
/tmp # java -jar arthas-boot.jar
[INFO] JAVA_HOME: /opt/java/openjdk
[INFO] arthas-boot version: 4.1.4
[INFO] Found existing java process, please choose one and input the serial number of the process, eg : 1. Then hit ENTER.
* [1]: 1 /app/app.jar
1
[INFO] arthas home: /root/.arthas/lib/4.1.5/arthas
[INFO] The target process already listen port 3658, skip attach.
[INFO] arthas-client connect 127.0.0.1 3658
,---. ,------. ,--------.,--. ,--. ,---. ,---.
/ O \ | .--. ''--. .--'| '--' | / O \ ' .-'
| .-. || '--'.' | | | .--. || .-. |`. `-.
| | | || |\ \ | | | | | || | | |.-' |
`--' `--'`--' '--' `--' `--' `--'`--' `--'`-----'
wiki https://arthas.aliyun.com/doc
tutorials https://arthas.aliyun.com/doc/arthas-tutorials.html
version 4.1.5
main_class /app/app.jar
pid 1
start_time 2026-01-15 11:13:33.854
current_time 2026-01-22 16:24:09.821
[arthas@1]$ dashboard -i 1000 -n 10000
浙公网安备 33010602011771号