English
Understanding the Report
A walk-through of every section the EasyGC analyzer produces.
Key indicators
| Metric | What it means | Healthy range |
|---|---|---|
| GC Throughput | % of total time the application (not the collector) runs | ≥ 95% |
| Avg Pause | Mean duration of stop-the-world collections | Depends on SLO; < 50 ms is comfortable |
| Max / P99 Pause | Worst and 99th-percentile pauses — the ones users feel | < 1 s |
| Young GC count | Number of minor collections | Normal to be frequent |
| Full GC count | Number of stop-the-world whole-heap collections | 0 — any Full GC deserves investigation |
| Total Pause Time | Sum of all stop-the-world time in the log | Compare against SLA budgets |
| Allocation Rate | Average object allocation speed (MB/s) | Watch after code changes |
| Log Time Span | Wall-clock / uptime range the log covers | Longer = more reliable |
JVM Memory
The maximum capacities observed in the log: total heap, young gen, old gen and metaspace. This tells you the effective -Xmx / -Xmn (or region) settings without touching the JVM.
TIP
Young/Old breakdown requires Java 8 style detailed logs. Unified logs (-Xlog:gc*) report whole-heap numbers; enable -Xlog:gc+heap* for region details.
Charts
- Heap Usage Around GC — the sawtooth of heap occupancy. The After GC line should oscillate in a stable band. If it climbs step by step, you are looking at retained objects that never get collected — a leak.
- GC Pause Times — every stop-the-world event over time, stacked by kind (young / full / CMS remark). Look for outliers and periodic spikes.
- GC Causes — what triggered each collection.
Allocation Failureis healthy;System.gc(),MetaspaceorPromotion failedare leads to follow.
Recommendations
Rules-based advice derived from your actual data, each with a severity:
- Critical — act now (memory leak, throughput below 50%).
- Warning — tune soon (Full GCs present, explicit GC calls, long max pause).
- Info — optimization opportunities (oversized heap, few events, humongous allocations).
- Good — patterns worth keeping (healthy throughput, ZGC sub-millisecond pauses).
GC Events
The full parsed event list: uptime, type, trigger cause, pause duration and heap sizes before/after every collection. Use it to correlate GC activity with your own application logs by timestamp.