Soak Testing

Soak testing (also called endurance testing) holds a realistic, sustained load on a system for a long period, typically hours to days. It looks for problems that only appear with time rather than with intensity.

Why it matters

Some defects are invisible in a ten-minute test. A memory leak of a few megabytes per hour, a connection that is never returned to the pool, a log file that slowly fills the disk, or a job that gets slightly slower each cycle will all pass a short load test and then cause an outage on day three of production.

Example

You run 200 requests per second against a staging build for 24 hours and graph these over time:

• Resident memory of each process (should be flat, not a rising line)
• Open file descriptors and database connections
• Disk usage and log volume
• Response time at the 95th percentile

If memory climbs from 400 MB to 1.8 GB across the day and p95 latency creeps up with it, you have a leak. A restart schedule can hide it, but the fix is finding the cause with a heap profile.

Common mistakes

• Using a constant, unrealistic request pattern. Include daily cycles and background jobs.
• Not recording baseline measurements, so you cannot tell normal growth from a leak.
• Forgetting that garbage collection can make memory look fine for hours before it fails.

Related

• Load testing
• Stress testing
• Saturation
• Metrics

यह पेज सामान्य जानकारी है, जिसे हमारी संपादकीय टीम ने लिखा है और दस्तावेज़ीकरण बदलने पर अपडेट किया जाता है। सर्वर और प्रदाता अलग-अलग होते हैं, इसलिए बदलाव पहले सुरक्षित तरीके से जाँचें। देखें: संपादकीय नीति और अस्वीकरण.