From b89c8099d8260d33d6428052adb27be1ea4296d3 Mon Sep 17 00:00:00 2001 From: Konstantin Renner Date: Fri, 10 Jul 2026 13:44:54 +0200 Subject: [PATCH] home-assistant: raise the memory limit from 8Gi to 16Gi The pod was OOMKilled (exit 137) on 2026-07-09 20:48:46 against the 8Gi limit. The 8Gi ceiling dates from when HA ran on the 12Gi k8s-edge VM; it now schedules on the general pool, where the nodes have ~47Gi, so there is no reason to keep it below the 16Gi the HAOS VM used to give this workload. The restart that OOMKill caused also broke roborock, dreame_vacuum and dwd_weather: a fresh container re-runs HA's uv-based requirement install, which upgraded cffi 2.0.0 -> 2.1.0 on disk after _cffi_backend 2.0.0 had already been imported, and every pycryptodome-backed integration then died on the resulting FFI version mismatch. Co-Authored-By: Claude Opus 4.8 --- kubernetes/apps/home-assistant/deployment.yaml | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/kubernetes/apps/home-assistant/deployment.yaml b/kubernetes/apps/home-assistant/deployment.yaml index 087d48b5..ff108fc5 100644 --- a/kubernetes/apps/home-assistant/deployment.yaml +++ b/kubernetes/apps/home-assistant/deployment.yaml @@ -66,13 +66,15 @@ spec: timeoutSeconds: 5 # ~10k entities. The legacy HAOS VM ran this workload with 16Gi; # the 4Gi limit OOM-looped (working set ~3.9Gi, tipped over on - # morning load spikes). 8Gi on the resized k8s-edge VM (12Gi). + # morning load spikes). 8Gi held on the edge VM but was OOMKilled + # again on 2026-07-09 now that the pod schedules on the general + # pool (~47Gi nodes), so it is back to the HAOS-era 16Gi. resources: requests: cpu: 500m memory: 2Gi limits: - memory: 8Gi + memory: 16Gi # No securityContext/privileged: the USB meter is gone (read via # MQTT now), which was the only reason HA needed privileged. # Reverse-proxy sidecar. ingress-nginx → :8124 → 127.0.0.1:8123. -- 2.54.0