Prune unused images on k3m1 too, and stop bulk removals timing out

Shipped

September 16, 2026 at 5:07 PM UTC

Author

Kamo

Commit

9bf02a7

k3m1 had no prune job, and kubelet image GC waits for 85% disk, so every deploy's SHA-tagged image piled up: 3,368 images (539G), 86 in use. Add a daily k3m1-image-prune alongside k1m1's weekly one. Both now pass --timeout 300s. crictl's 2s default made 92-131 removals per k1m1 run fail with DeadlineExceeded.