From 686c962ea9a498500b13bd535daea706a015e57c Mon Sep 17 00:00:00 2001
From: rock <locartrock@gmail.com>
Date: Sun, 30 Aug 2026 09:23:09 -0700
Subject: [PATCH] fix(ci): aggressive GC for heavy Rust cargo builds

- GC CronJob runs 2x daily (02:00 & 05:00 UTC) instead of once (03:00)
- Delete incomplete actcache uploads immediately (tmp/ dirs from failed writes)
- Reduce actcache retention from 3 days to 1 day
- Prevents cargo cache backlog on rust runner under heavy commit load
- Incomplete entries were accumulating 900MB+ each, filling 1Gi PVC instantly
---
 .../forgejo-runner/templates/gc-cronjob.yaml   | 18 ++++++++++++------
 k8s/infra/forgejo-runner/values.yaml           |  2 +-
 2 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/k8s/infra/forgejo-runner/templates/gc-cronjob.yaml b/k8s/infra/forgejo-runner/templates/gc-cronjob.yaml
index 79fb29d..bed8ce8 100644
--- a/k8s/infra/forgejo-runner/templates/gc-cronjob.yaml
+++ b/k8s/infra/forgejo-runner/templates/gc-cronjob.yaml
@@ -43,7 +43,7 @@ metadata:
   labels:
     app: forgejo-runner-gc
 spec:
-  schedule: {{ .Values.gc.schedule | quote }}
+  schedule: "0 2,5 * * *"      # Run at 02:00 and 05:00 UTC (more frequent for heavy build workloads)
   concurrencyPolicy: Forbid
   successfulJobsHistoryLimit: 3
   failedJobsHistoryLimit: 3
@@ -126,12 +126,18 @@ spec:
                     echo "[docker] after:"
                     kubectl -n {{ .Release.Namespace }} exec "$POD" -c dind -- docker system df 2>/dev/null || true
 
-                    # 2. Actcache cleanup (runner container) — remove cached artifacts older than pruneAge
-                    echo "[actcache] cleaning stale cache entries..."
+                    # 2. Actcache cleanup (runner container)
+                    echo "[actcache] cleaning incomplete and stale cache entries..."
                     kubectl -n {{ .Release.Namespace }} exec "$POD" -c runner -- \
-                      sh -c 'find /data/.cache/actcache/cache -type f -mtime +{{ .Values.gc.actcacheMaxAgeDays }} -delete 2>/dev/null; \
-                             find /data/.cache/actcache/cache -type d -empty -delete 2>/dev/null; \
-                             echo "actcache size: $(du -sh /data/.cache/actcache/cache 2>/dev/null | cut -f1)"' || true
+                      sh -c '
+                        # Delete incomplete/partial cache uploads immediately (tmp dirs)
+                        find /data/.cache/actcache/cache -name "tmp" -type d -exec rm -rf {} + 2>/dev/null || true
+                        # Delete cache entries not accessed in last {{ .Values.gc.actcacheMaxAgeDays }} day(s)
+                        find /data/.cache/actcache/cache -type f -mtime +{{ .Values.gc.actcacheMaxAgeDays }} -delete 2>/dev/null || true
+                        # Clean up empty directories
+                        find /data/.cache/actcache/cache -type d -empty -delete 2>/dev/null || true
+                        echo "actcache size: $(du -sh /data/.cache/actcache/cache 2>/dev/null | cut -f1)"
+                      ' || true
 
                     echo ""
                   done
diff --git a/k8s/infra/forgejo-runner/values.yaml b/k8s/infra/forgejo-runner/values.yaml
index c7cd5f5..4ef52d0 100644
--- a/k8s/infra/forgejo-runner/values.yaml
+++ b/k8s/infra/forgejo-runner/values.yaml
@@ -63,4 +63,4 @@ gc:
   image: alpine/k8s:1.31.0
   pruneAge: "72h"              # Docker artifacts unused longer than this get pruned
   pruneAgeHours: 72             # Same as pruneAge but numeric for date arithmetic in shell
-  actcacheMaxAgeDays: 3        # actcache files older than N days get deleted
+  actcacheMaxAgeDays: 1        # actcache files older than N days (aggressive for heavy Rust cargo builds)