DongHyeonkaandClaude Opus 5
99b689e715
Keycloak commits the login INSERT with synchronous_commit off, so a crash loses whole sessions and not just refresh timestamps. Measured 4 of 153 lost, matching the default wal_writer_delay window. Two injections failed silently first: --grace-period=0 --force lets the container runtime send SIGTERM so PostgreSQL flushes and shuts down cleanly, and SIGKILL to PID 1 from inside its own namespace is ignored by the kernel. Killing a backend makes the postmaster reinitialize, which is a real crash recovery. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
18 lines
796 B
Plaintext
18 lines
796 B
Plaintext
=== [1] 로그인 루프 시작 (호스트에서 백그라운드로 exec — 세션이 살아 있어야 한다) ===
|
|
8초 동안 클라이언트가 200 을 받은 로그인: 106 건
|
|
|
|
=== [2] SIGKILL ===
|
|
종료: 12:01:32.981
|
|
반환: 12:01:33.236
|
|
최종 성공 로그인 수: 110 건
|
|
마지막 sid: FimM-krSybBACP2qIvshLWwU
|
|
마지막 sid: EwFfFwOIfqiv8N5GQ5OjtsVq
|
|
마지막 sid: CJX-PxFkS7rUc_9FQgB7iw1f
|
|
마지막 sid: 1EFK7SgUA4M7tq_SkC_BD2er
|
|
마지막 sid: _LiqTczuyxlpOs3T3xs25SLv
|
|
|
|
=== [3] PostgreSQL 재기동 후 crash recovery 확인 ===
|
|
Waiting for deployment "postgres" rollout to finish: 0 of 1 updated replicas are available...
|
|
deployment "postgres" successfully rolled out
|
|
2026-09-04 02:59:48.427 UTC [1] LOG: database system is ready to accept connections
|