عرض عملي: استنزاف المجمّع تحت الحِمل

خطوة بخطوة

1. Run Shop with pool size 5 and a load test of 100 concurrent users on checkout.
2. Watch metrics: hikaricp_connections_active at max, pending growing, latencies climbing, then timeout errors.
3. Take a thread dump and find threads waiting in getConnection and the holders inside a slow remote call.
4. Planned mistake: raise the pool to 200 to "fix" it; database CPU saturates and latency does not improve. Revert.
5. Real fix: move the remote call out of the transaction; reduce hold time; right-size the pool.
6. Re-run the load test and compare.

Prepares MID-017-LAB1.

اختبر فهمك