From 0ff84ab3cb44421a9c3a51addb2bf93cb533ba04 Mon Sep 17 00:00:00 2001 From: kenzycodex Date: Wed, 11 Nov 2026 08:18:16 +0000 Subject: [PATCH 1/2] fix(deploy): the site reserves 128Mi, so its update fits next to the API in the quota --- deploy/forge/app.yaml | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/deploy/forge/app.yaml b/deploy/forge/app.yaml index 629da93..8315126 100644 --- a/deploy/forge/app.yaml +++ b/deploy/forge/app.yaml @@ -13,7 +13,8 @@ # Secrets are never written here. The team key comes from the `huawei-gateway` # secret the platform provisioned (as GATEWAY_KEY); Fable's own settings come # from a `fable-env` secret created once (docs/DEPLOY-FORGE.md). -# Quota check: requests 200m / 512Mi of 400m / 768Mi, limits 1 core / 1Gi of +# Quota check: requests 200m / 384Mi of 400m / 768Mi (512Mi while the site +# updates, when two of its pods overlap), limits 1 core / 1Gi of # 2 cores / 2Gi, 2 pods of 10, 2 Services of 5, no persistent volume. apiVersion: apps/v1 kind: Deployment @@ -130,7 +131,9 @@ spec: initialDelaySeconds: 20 periodSeconds: 20 resources: - requests: { cpu: 100m, memory: 256Mi } + # 128Mi reserved: while the site updates, two copies run side by side + # next to the API (256 + 128 + 128 of the namespace's 768Mi). + requests: { cpu: 100m, memory: 128Mi } limits: { cpu: 500m, memory: 512Mi } --- apiVersion: v1 From c137763c07f3718cd96c79ab9f3dba66f9f9d73a Mon Sep 17 00:00:00 2001 From: kenzycodex Date: Wed, 11 Nov 2026 08:18:16 +0000 Subject: [PATCH 2/2] docs(deploy): what to do when the API pod is refused by the quota after an update --- docs/DEPLOY-FORGE.md | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/DEPLOY-FORGE.md b/docs/DEPLOY-FORGE.md index df88961..88fd76f 100644 --- a/docs/DEPLOY-FORGE.md +++ b/docs/DEPLOY-FORGE.md @@ -195,6 +195,7 @@ kubectl rollout undo deployment/app | --- | --- | --- | | `ImagePullBackOff` | `regcred` missing or its token expired | Recreate the secret | | Pod `Pending`, "exceeded quota" | Requests above the namespace quota | `kubectl describe resourcequota team-quota` | +| No API pod after an update; events say "exceeded quota" | The quota still counts pods that just stopped | Wait until `requests.memory` drops, then `kubectl rollout restart deployment/api` | | Pod never Ready | The process did not start | `kubectl logs deploy/api --previous` | | `api` exits at start naming unsafe settings | `fable-env` is missing a value | Recreate the secret from "Once" | | `/health` says the database is not writable | The data directory is not writable by uid 10001 | Check `fsGroup: 10001` is still in the manifest |