testland/jvm-gc-tuning
Diagnoses JVM garbage-collection behaviour under load: reads and interprets unified GC logs (-Xlog:gc*), selects the right collector (G1 vs ZGC vs Parallel vs Serial), tunes heap sizing and pause-time targets, quantifies allocation rate, and traces the GC-pause-to-latency-tail link using GCViewer and Java Flight Recorder (JFR). Use when a load test reveals p99/p999 latency spikes that correlate with GC activity, or when heap sizing and collector selection need justification before a performance baseline is locked.
74
93%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
View guide
Passed
No findings from the security scan
gc-reference.mdreferences/
JVM GC reference tables
Lookup tables for jvm-gc-tuning: G1 tuning flags, JFR event categories, and
the metrics GCViewer surfaces. The runnable capture and analysis commands live
in the main SKILL.md; these are the reference details each step points to.
Key G1 flags (Step 4)
From the G1 GC Tuning section:
| Flag | Default | Effect |
|---|---|---|
-XX:MaxGCPauseMillis | 200 | Pause-time target; G1 sizes young gen to meet this with high probability |
-XX:InitiatingHeapOccupancyPercent | 45% | Triggers concurrent marking when old gen occupancy reaches this level |
-XX:G1HeapRegionSize | ergonomic (~2048 regions) | Region size 1-512 MB (power of 2); objects >= half this become humongous |
-XX:G1MixedGCLiveThresholdPercent | 85% | Regions with live occupancy above this are skipped during mixed GC |
-XX:G1HeapWastePercent | 5% | Space-reclamation phase ends when reclaimable space drops below this |
G1 is not a real-time collector; the pause target is a statistical goal. Setting it below 10 ms without switching to ZGC typically causes G1 to thrash with very frequent tiny collections.
Key JFR event categories (Step 7)
| Event category | What to look for |
|---|---|
jdk.GarbageCollection | Duration, cause, GC ID per collection |
jdk.G1HeapSummary | Eden/survivor/old occupancy before and after |
jdk.ObjectAllocationInNewTLAB | Allocation call stacks (sampled) |
jdk.ObjectAllocationOutsideTLAB | Large object allocations (humongous) |
jdk.GCHeapSummary | Heap used/committed at collection start |
Key GCViewer metrics (Step 8)
| Metric | GC health interpretation |
|---|---|
| Throughput % | % of time NOT in GC; target > 95% for most services |
| Avg pause time | Mean STW duration; should be below MaxGCPauseMillis |
| Max pause time | Worst-case STW; the main driver of p99/p999 latency |
| Allocation rate (MB/s) | Matches the Step 6 allocation-rate analysis |
| Promotion rate (MB/s) | Objects escaping young gen; high rate fills old gen faster |