CODEWITHSUNDEEP
javag1gc
hongyiheng
hongyiheng

Reputation: 3

G1 GC take a long time for Object Copy

My application is a typical Java web application, Java 8, using G1 GC. I have encountered a GC problem. Within 10 minutes after the application is started (when the old generation is gradually increasing), long GC often occurs, but this does not happen during the stable operation stage. The following is my GC log:

{Heap before GC invocations=43 (full 0):
 garbage-first heap   total 5242880K, used 496183K [0x00000006a7800000, 0x00000006a7a05000, 0x00000007e7800000)
  region size 2048K, 128 young (262144K), 54 survivors (110592K)
 Metaspace       used 228219K, capacity 255729K, committed 255872K, reserved 1275904K
  class space    used 24456K, capacity 29030K, committed 29056K, reserved 1048576K
2024-06-20T11:44:47.168+0800: 260.143: [GC pause (G1 Evacuation Pause) (young), 9.8956459 secs]
   [Parallel Time: 9891.3 ms, GC Workers: 4]
      [GC Worker Start (ms): Min: 260143.9, Avg: 260145.0, Max: 260147.2, Diff: 3.3]
      [Ext Root Scanning (ms): Min: 5.1, Avg: 8.8, Max: 12.5, Diff: 7.5, Sum: 35.2]
      [Update RS (ms): Min: 0.0, Avg: 12.2, Max: 16.2, Diff: 16.2, Sum: 48.6]
         [Processed Buffers: Min: 0, Avg: 39.2, Max: 69, Diff: 69, Sum: 157]
      [Scan RS (ms): Min: 0.0, Avg: 1.1, Max: 1.9, Diff: 1.8, Sum: 4.4]
      [Code Root Scanning (ms): Min: 0.0, Avg: 4.0, Max: 6.1, Diff: 6.1, Sum: 15.8]
      [Object Copy (ms): Min: 9855.0, Avg: 9863.9, Max: 9884.8, Diff: 29.8, Sum: 39455.6]
      [Termination (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0]
         [Termination Attempts: Min: 1, Avg: 15.8, Max: 23, Diff: 22, Sum: 63]
      [GC Worker Other (ms): Min: 0.1, Avg: 0.1, Max: 0.1, Diff: 0.0, Sum: 0.2]
      [GC Worker Total (ms): Min: 9887.8, Avg: 9890.0, Max: 9891.1, Diff: 3.3, Sum: 39560.0]
      [GC Worker End (ms): Min: 270035.0, Avg: 270035.0, Max: 270035.0, Diff: 0.1]
   [Code Root Fixup: 1.1 ms]
   [Code Root Purge: 0.0 ms]
   [Clear CT: 0.2 ms]
   [Other: 3.0 ms]
      [Choose CSet: 0.0 ms]
      [Ref Proc: 1.1 ms]
      [Ref Enq: 0.0 ms]
      [Redirty Cards: 0.5 ms]
      [Humongous Register: 0.1 ms]
      [Humongous Reclaim: 0.0 ms]
      [Free CSet: 0.4 ms]
   [Eden: 148.0M(148.0M)->0.0B(234.0M) Survivors: 108.0M->22.0M Heap: 484.6M(5120.0M)->338.4M(5120.0M)]
Heap after GC invocations=44 (full 0):
 garbage-first heap   total 5242880K, used 346567K [0x00000006a7800000, 0x00000006a7a05000, 0x00000007e7800000)
  region size 2048K, 11 young (22528K), 11 survivors (22528K)
 Metaspace       used 228219K, capacity 255729K, committed 255872K, reserved 1275904K
  class space    used 24456K, capacity 29030K, committed 29056K, reserved 1048576K
}
 [Times: user=17.10 sys=0.90, real=9.90 secs]

I tried increasing the SURVIVOR area, but it didn't work. Here are my JVM parameters:

-Xmx5120m -Xms5120m -XX:MaxMetaspaceSize=400M -XX:MetaspaceSize=400M -XX:+UseG1GC -XX:SurvivorRatio=6 -Dnacos.use.endpoint.parsing.rule=false -Dnacos.use.cloud.namespace.parsing=false -XX:+PrintHeapAtGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+PrintGCApplicationStoppedTime -XX:+PrintGCApplicationConcurrentTime -Xloggc:/tmp1/main-gc.log

Upvotes: 0

Views: 134

Answers (1)

Munkh-Erdene Jargal
Munkh-Erdene Jargal

Reputation: 101

The issue may stem from other aspects of the G1 garbage collection configuration or the application's memory usage patterns. Here are additional strategies and adjustments you can try to mitigate the long GC pauses:

-XX:G1HeapRegionSize=4M
-XX:MaxGCPauseMillis=200
-XX:G1NewSizePercent=30 -XX:G1MaxNewSizePercent=60
-XX:ParallelGCThreads=8 -XX:ConcGCThreads=4
-XX:+UseStringDeduplication

Further things to do

  1. Heap Dump Analysis: Capture and analyze heap dumps during the problematic phase to understand what objects are consuming memory. Utilize profiling tools like VisualVM, YourKit, or Java Mission Control to analyze the memory usage patterns and object allocations. My favorite is JProfiler
  2. Monitor GC Logs: Continuously monitor and analyze the GC logs to identify any patterns or specific allocations that might be causing the long GC pauses. Look for frequent or long G1 Evacuation Pauses
  3. Gradual Increase: If applicable, gradually increase the heap size to give more room for objects, allowing the GC to work more efficiently.

Upvotes: 0

Related Questions