| commit | 319a435db60ce4a3233cee79d18e31df8021288d | [log] [tgz] |
|---|---|---|
| author | Nathan Mulcahey <nmulcahey@google.com> | Mon Apr 19 18:13:19 2021 +0000 |
| committer | Nathan Mulcahey <nmulcahey@google.com> | Mon Apr 19 18:13:19 2021 +0000 |
| tree | ab439ff3cddf0fb001011e8d2fc58f8e9fb1691e | |
| parent | 6733b283d7a2748c84e3a0f0d25581bf785e9381 [diff] |
[exec-server] Throttle scaleUp. This limits scaleUp operations to 10 Pods every 5 seconds up to the maximum desired value that has been recorded over the last 300 seconds. Previously, it used the default which is to double every 15 seconds based on the currently desired value. Bug: 74649 Change-Id: I229f06890df414843888c240acdd5fb26b8a2e2c Reviewed-on: https://fuchsia-review.googlesource.com/c/infra/config/+/516326 Reviewed-by: Haowei Wu <haowei@google.com>
diff --git a/goma/k8s/templates-rbe/goma/hpa_exec-server.yaml.in b/goma/k8s/templates-rbe/goma/hpa_exec-server.yaml.in index 14c118c..f112709 100644 --- a/goma/k8s/templates-rbe/goma/hpa_exec-server.yaml.in +++ b/goma/k8s/templates-rbe/goma/hpa_exec-server.yaml.in
@@ -3,6 +3,13 @@ metadata: name: exec-server spec: + behavior: + scaleUp: + policies: + - type: Pods + value: 10 + periodSeconds: 5 + stabilizationWindowSeconds: 300 maxReplicas: 320 minReplicas: 16 scaleTargetRef: