103 lines
3.9 KiB
YAML
103 lines
3.9 KiB
YAML
apiVersion: apps/v1
|
|
kind: Deployment
|
|
metadata:
|
|
name: hrrr-point-rs
|
|
namespace: prop
|
|
# Phase 3 Stream C: per-QSO HRRR fetches move off the BEAM pods.
|
|
# Elixir writes to hrrr_fetch_tasks (one row per valid_time with a
|
|
# points JSONB array); this deployment drains via FOR UPDATE SKIP
|
|
# LOCKED, fetches each HRRR cycle once, and upserts hrrr_profiles.
|
|
spec:
|
|
# Parallelism on QSO-submit spikes comes from multiple replicas
|
|
# claiming rows from hrrr_fetch_tasks via FOR UPDATE SKIP LOCKED —
|
|
# no coordination needed between workers.
|
|
# Replica count is HPA-managed (hpa.yaml, min 1 / max 4).
|
|
# Zero-downtime rollout: surge first so the new pod is serving
|
|
# before the old one drains, otherwise a deploy during a QSO-submit
|
|
# burst leaves the per-QSO fetch lane empty until the new pod is
|
|
# ready. Both pods race for hrrr_fetch_tasks via FOR UPDATE SKIP
|
|
# LOCKED, so brief overlap is safe.
|
|
strategy:
|
|
type: RollingUpdate
|
|
rollingUpdate:
|
|
maxSurge: 1
|
|
maxUnavailable: 0
|
|
selector:
|
|
matchLabels:
|
|
app: hrrr-point-rs
|
|
template:
|
|
metadata:
|
|
labels:
|
|
app: hrrr-point-rs
|
|
tier: hrrr-points
|
|
spec:
|
|
# Anti-affinity against prop-grid-rs was removed; the scheduler
|
|
# is free to colocate. Contention on wgrib2 + NFS hasn't been an
|
|
# observed bottleneck in practice.
|
|
# Per-QSO point fetches are short (~5-15 s), but a stack of
|
|
# claimed rows can take a couple of minutes to drain. 120 s lets
|
|
# the worker finish in-flight tasks before SIGKILL.
|
|
terminationGracePeriodSeconds: 120
|
|
imagePullSecrets:
|
|
- name: forgejo-registry
|
|
securityContext:
|
|
runAsUser: 65534
|
|
runAsNonRoot: true
|
|
fsGroup: 65534
|
|
seccompProfile:
|
|
type: RuntimeDefault
|
|
containers:
|
|
- name: hrrr-point-rs
|
|
# Shares the prop-grid-rs image so one CI pipeline produces
|
|
# both binaries. k8s `command:` overrides the default
|
|
# ENTRYPOINT to launch the per-QSO point worker.
|
|
image: git.mcintire.me/graham/prop-grid-rs:main-1777132414-6aa91e7 # {"$imagepolicy": "flux-system:prop-grid-rs"}
|
|
imagePullPolicy: IfNotPresent
|
|
command: ["/usr/bin/tini", "--", "/usr/local/bin/hrrr-point-worker"]
|
|
env:
|
|
- name: HRRR_BASE_URL
|
|
value: "http://skippy.w5isp.com:8080"
|
|
- name: HRRR_FALLBACK_BASE_URL
|
|
value: "https://noaa-hrrr-bdp-pds.s3.amazonaws.com"
|
|
- name: RUST_LOG
|
|
value: "info"
|
|
- name: PROP_GRID_RS_PG_CONNS
|
|
value: "4"
|
|
# Shared NFS idx cache — same directory as prop-grid-rs so a
|
|
# chain-step idx fetch warms both workers.
|
|
- name: HRRR_IDX_CACHE_DIR
|
|
value: "/data/hrrr_idx"
|
|
envFrom:
|
|
- secretRef:
|
|
name: prop-secrets
|
|
resources:
|
|
requests:
|
|
cpu: 100m
|
|
memory: 256Mi
|
|
limits:
|
|
# Per-batch peak is a single CONUS surface + pressure
|
|
# decode (~40 MB blob, ~200 MB wgrib2 working set plus
|
|
# the full 92k-cell merged CellValues map kept in memory
|
|
# until all requested points are drained). 1 Gi OOMKilled
|
|
# repeatedly — four restarts in ten minutes — so bumped
|
|
# to 2 Gi. The scheduler-on-worker2 placement has plenty
|
|
# of headroom.
|
|
cpu: "2"
|
|
memory: 2Gi
|
|
securityContext:
|
|
allowPrivilegeEscalation: false
|
|
runAsNonRoot: true
|
|
runAsUser: 65534
|
|
capabilities:
|
|
drop:
|
|
- ALL
|
|
seccompProfile:
|
|
type: RuntimeDefault
|
|
volumeMounts:
|
|
- name: data
|
|
mountPath: /data
|
|
volumes:
|
|
- name: data
|
|
nfs:
|
|
server: 10.0.15.103
|
|
path: /data
|