Files
aevolution/deploy/kubernetes/region-server/deployment.yaml.example
T
dingoandClaude Opus 5.5 7aec68db8e Complete M7.8 operations endpoints and region server image
ADR-0066: evo-server --ops-listen serves /livez (loop responding, never
authority), /readyz (200 only while the host holds the lease with a
verified recovered head) and /metrics (Prometheus text: 30 fixed
series for readiness, phase, lease epoch/remaining, simulated/committed
ticks, commit lag, checkpoint age, tick debt, queue use and bounds, and
counters; the only labels are a 6-value phase and 4-value drop reason).

The dependency-free OpsServer runs on the host loop with no thread:
GET on exact paths, 405/404/400/431 responses, 8 connections with excess
refused on accept, 4 KiB requests, 1 s deadlines, 16 KiB responses.
HostStatus/phase types moved to the portable evo::ops library with a
Windows-tested renderer. evo-server now refuses to start without its
required identity and connection settings.

Adds the server-runtime Docker target (evo-server + pinned libpq,
UID 10001, no credentials), a CI server-image job with smoke script and
optional publishing, and a Kubernetes example outside Argo.

Validated: Windows CTest 59/59; Linux Debug/Release/clang-tidy/ASan
CTest 59/59; PostgreSQL suites under Release, clang-tidy and ASan with
real process probes; two ops mutations caught. Docker build/smoke and
Gitea CI not executed locally.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-28 23:09:28 -04:00

144 lines
5.6 KiB
Plaintext

# M7.8 development region server (ADR-0066). NOT in the active Argo kustomization.
# Before applying to evolution-dev: pin the server image digest, provision the region
# (owner-only), create the runtime Secret and object PVC, and add DB/DNS egress for
# the real endpoint. Never target World0 from this example.
#
# Authority, not liveness, gates traffic: /readyz is 200 only while this pod holds
# the region lease and has recovered and verified the committed head. /livez only
# says the loop is responding. A probe can wait up to one bounded I/O step (commit
# deadline 5 s), so probe timeouts below exceed it.
#
# There is no gateway yet: stdin is not attached, so the region hosts authority but
# receives no inputs (tick debt grows by design; nothing is invented).
apiVersion: apps/v1
kind: Deployment
metadata:
name: evolution-region-7-1
namespace: evolution-dev
labels: {app.kubernetes.io/name: evolution-region, evolution.aeternalis/region: "7-1"}
spec:
replicas: 1
strategy: {type: Recreate} # Never two pods racing; the lease still arbitrates.
selector:
matchLabels: {app.kubernetes.io/name: evolution-region, evolution.aeternalis/region: "7-1"}
template:
metadata:
labels: {app.kubernetes.io/name: evolution-region, evolution.aeternalis/region: "7-1"}
spec:
automountServiceAccountToken: false
terminationGracePeriodSeconds: 30 # SIGTERM drains within --drain-ms 20000.
securityContext:
runAsNonRoot: true
runAsUser: 10001
runAsGroup: 10001
fsGroup: 10001
seccompProfile: {type: RuntimeDefault}
initContainers:
# libpq rejects group-readable password files: copy to a private 0600 file.
- name: pgpass
image: registry.aeternalisgames.com/aeternalis/aevolution-server:replace-with-immutable-digest
command: ["/bin/sh", "-c", "install -m 0600 /run/postgres/pgpass /run/evo/pgpass"]
securityContext: {readOnlyRootFilesystem: true, allowPrivilegeEscalation: false, capabilities: {drop: [ALL]}}
volumeMounts:
- {name: connection, mountPath: /run/postgres, readOnly: true}
- {name: private, mountPath: /run/evo}
containers:
- name: region
image: registry.aeternalisgames.com/aeternalis/aevolution-server:replace-with-immutable-digest
# Flags are "--name value" pairs; see docs/infrastructure/POSTGRESQL.md.
command:
- /usr/local/bin/evo-server
- --profile
- lineage
- --world
- "7"
- --region
- "1"
- --frame
- "9"
- --pg-host
- replace-with-tls-hostname
- --pg-hostaddr
- replace-with-numeric-address
- --pg-port
- "5432"
- --pg-database
- evolution
- --pg-user
- replace-with-runtime-login
- --pg-passfile
- /run/evo/pgpass
- --pg-ca
- /run/postgres/ca.crt
- --objects
- /var/lib/evolution/objects
- --build
- replace-with-retained-build-identity-hex
- --ops-listen
- 0.0.0.0:9100
- --drain-ms
- "20000"
ports:
- {name: ops, containerPort: 9100}
startupProbe:
httpGet: {path: /livez, port: ops}
periodSeconds: 5
timeoutSeconds: 8
failureThreshold: 12
readinessProbe:
httpGet: {path: /readyz, port: ops}
periodSeconds: 5
timeoutSeconds: 8
failureThreshold: 1 # Any doubt about authority removes the pod at once.
livenessProbe:
httpGet: {path: /livez, port: ops}
periodSeconds: 10
timeoutSeconds: 8
failureThreshold: 6
resources:
requests: {cpu: 250m, memory: 128Mi}
limits: {cpu: "1", memory: 512Mi}
securityContext:
readOnlyRootFilesystem: true
allowPrivilegeEscalation: false
capabilities: {drop: [ALL]}
volumeMounts:
- {name: connection, mountPath: /run/postgres, readOnly: true}
- {name: private, mountPath: /run/evo, readOnly: true}
- {name: objects, mountPath: /var/lib/evolution/objects}
- {name: scratch, mountPath: /tmp}
volumes:
- name: connection
secret:
secretName: evolution-postgres-runtime # Runtime login only; never migrator.
defaultMode: 0440
items:
- {key: pgpass, path: pgpass}
- {key: ca.crt, path: ca.crt}
- name: private
emptyDir: {medium: Memory, sizeLimit: 1Mi}
- name: objects
persistentVolumeClaim: {claimName: evolution-region-objects}
- name: scratch
emptyDir: {medium: Memory, sizeLimit: 8Mi}
---
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: evolution-region-7-1
namespace: evolution-dev
spec:
podSelector:
matchLabels: {app.kubernetes.io/name: evolution-region, evolution.aeternalis/region: "7-1"}
policyTypes: [Ingress, Egress]
ingress:
# Metrics/probes only from the monitoring namespace (kubelet probes use the node).
- from:
- namespaceSelector:
matchLabels: {kubernetes.io/metadata.name: monitoring}
ports: [{port: 9100, protocol: TCP}]
egress:
# Replace with the actual database endpoint; no DNS is needed with a numeric hostaddr.
- to: [{ipBlock: {cidr: 192.0.2.10/32}}]
ports: [{port: 5432, protocol: TCP}]