Commit Graph
9 Commits
Author SHA1 Message Date
DongHyeonkaandClaude Opus 5 4177fb6a48 docs: A-2 — losing the database takes every node down while up stays 1
Both pods go NotReady, the Service endpoint list empties and the front door returns 503, so adding Keycloak replicas buys nothing against database loss. The node holding the session in cache fails too, because a refresh writes LAST_SESSION_REFRESH. Recovery was automatic in about fifteen seconds with no restart, which is what readiness rather than liveness buys.

The observability finding matters as much: up stayed at 1 through a total outage, so alerting on it would have caught nothing. kube-state-metrics is missing and pod readiness is therefore not recorded as a metric.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-04 11:57:55 +09:00
DongHyeonkaandClaude Opus 5 2a98ef1090 docs: A-1 — sessions survive a JGroups partition but logout invalidation does not
Cutting TCP 7800 leaves cross-node refresh working (200), confirming sessions travel through PostgreSQL rather than the cluster transport. Logout is the opposite: the database row is deleted but the other node answers from its stale local cache, so the A-0 conclusion that invalidation rides the database is corrected here.

Two things the plan did not anticipate: a NetworkPolicy cannot sever an established connection because conntrack accepts it before policy evaluation, and Keycloak reports the partition through its readiness probe so the split node removes itself from the Service.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-04 11:52:49 +09:00
DongHyeonkaandClaude Opus 5 22d873eb4f docs: capture the SQL the other node actually runs, and correct the replication claim
PostgreSQL statement logging shows keycloak-1 reading and updating the session created on keycloak-0. The same transaction reveals optimistic locking via VERSION, SKIP LOCKED, and synchronous_commit turned off. Fixes the earlier concept note that credited Infinispan with cross-node propagation.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-04 10:14:45 +09:00
DongHyeonkaandClaude Opus 5 e5ebaeb623 docs: prove sessions are shared by PostgreSQL, not Infinispan replication
Experiment 0 with three probes: cross-node refresh/logout, cache counter deltas around a single login, and cache entry ownership. Each node caches only what it handled; cache totals sum exactly to the database count.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-04 09:58:10 +09:00
DongHyeonkaandClaude Opus 5 33878e8880 docs: record multi-node cluster setup, rationale and formation evidence
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-03 17:26:45 +09:00
DongHyeonkaandClaude Opus 5 e1ba9c5626 docs: record proxy-bypass closure with before and after evidence
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-03 16:17:24 +09:00
DongHyeonkaandClaude Opus 5 98874b0c6c docs: retake staged evidence screenshots with indented output
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-03 15:52:47 +09:00
DongHyeonkaandClaude Opus 5 7737787937 docs: record the staged fix and post-fix evidence
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-03 15:33:40 +09:00
DongHyeonkaandClaude Opus 5 69d4502757 docs: collect evidence for the two-hop header contract failure
Control experiment isolates two independent causes: the nginx 443 block still emits X-Forwarded-Proto http, and Traefik rewrites forwarded headers regardless.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-03 15:04:38 +09:00