release-tests/tekton/pvc/go-module-cache.yaml at d384d2072978bcdd7a707ec8e9780734475ffce8 · openshift-pipelines/release-tests · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
# go-module-cache PVC
# ─────────────────────────────────────────────────────────────────────────────
# Shared, persistent PVC for the Go module download cache.
#
# Why ReadWriteMany?
#   Multiple release-tests TaskRun pods run in parallel (one per test suite).
#   All of them need to mount the same volume concurrently - that requires an
#   RWX access mode.  On ROSA/OSD/OCP the default StorageClass is usually
#   backed by Ceph RBD (RWO only).  Ensure you use a StorageClass that supports
#   ReadWriteMany (e.g. ocs-storagecluster-cephfs, nfs, or azurefile).
#
# Lifecycle:
#   - This PVC is long-lived (not recreated per PipelineRun).
#   - It is populated once by the warm-go-cache Task before parallel test tasks
#     start.  Subsequent runs reuse the existing cache; only changed modules are
#     fetched.
#   - If go.mod / go.sum change significantly, delete and recreate the PVC to
#     avoid stale entries (or let warm-go-cache handle it with go mod download,
#     which is additive-only).
#
# Apply once per cluster with:
#   oc apply -f tekton/pvc/go-module-cache.yaml -n openshift-pipelines
# ─────────────────────────────────────────────────────────────────────────────
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: go-module-cache
  namespace: openshift-pipelines
  labels:
    app.kubernetes.io/part-of: release-tests
    app.kubernetes.io/component: go-cache
  annotations:
    # Human-readable explanation of why this PVC exists
    release-tests.tekton.dev/purpose: >
      Shared Go module cache populated once per pipeline run by the
      warm-go-cache task, then mounted read-only by all parallel
      release-tests tasks to eliminate concurrent module downloads that
      cause gauge runner timeout (>101 min).
spec:
  accessModes:
    # ReadWriteMany is required so all parallel test pods can mount
    # simultaneously.  Use ocs-storagecluster-cephfs or equivalent.
    - ReadWriteMany
  resources:
    requests:
      # The full Go module graph for this project is ~1.5 GB.
      # 5 Gi gives comfortable headroom for future growth.
      storage: 5Gi
  # Replace with the RWX-capable StorageClass available on your cluster.
  # Examples:
  #   OCP/OSD/ROSA (OCS)   : ocs-storagecluster-cephfs
  #   ARO                  : azurefile-csi
  #   GKE                  : standard-rwx
  #   Generic NFS          : nfs-client
  storageClassName: ocs-storagecluster-cephfs