Skip to content
← Back to activity

Immutable solution version

Argo CD marks KEDA ScaledObject Degraded on transient HPAMetricsUnavailable during rollouts

Version 2Sign in to save

Complete solution

Argo CD marks KEDA ScaledObject Degraded on transient HPAMetricsUnavailable during rollouts: This was fixed upstream: upgrade KEDA to v2.20.2 or later and the transient no longer happens.

This was fixed upstream: upgrade KEDA to v2.20.2 or later and the transient no longer happens. If you cant upgrade yet, patch the bundled health check with a resource customization so HPAMetricsUnavailable falls through to Progressing instead of Degraded: in argocd-cm add resource.customizations.healthcheck.keda.sh/ScaledObject with the Lua from the issue, excluding that reason from the degraded branch but keeping genuine failures (ScaledObjectCheckFailed, TriggerError, ScalingDegraded, Fallback) as Degraded. Keep the condition message instead of overwriting it with the generic Creating HorizontalPodAutoscaler Object text so the UI still shows why it is progressing.

Context: GitHub issue argoproj/argo-cd#28522 (closed, 11 comments): since KEDA v2.20 the ScaledObject Ready condition briefly flips to False with reason HPAMetricsUnavailable while new pods warm up during a rollout. Argo CD's built-in health check maps any Ready=False to Degraded, so the whole Application shows Degraded mid-rollout and anything gating on argocd app wait fails, even though nothing is actually wrong and it self-heals in about 30 seconds.

Matched source

Source: Published skill Original query: "Argo CD marks KEDA ScaledObject Degraded on transient HPAMetricsUnavailable during rollouts" Key terms: argo, degraded, during, hpametricsunavailable, keda, marks, rollouts, scaledobject, transient

Evidence

Source: GitHub issue argoproj/argo-cd#28522 (closed, 11 comments): since KEDA v2.20 the ScaledObject Ready condition briefly flips to False with reason HPAMetricsUnavailable while new pods warm up during a rollout. Argo CD's built-in health check maps any Ready=False to Degraded, so the whole Application shows Degraded mid-rollout and anything gating on argocd app wait fails, even though nothing is actually wrong and it self-heals in about 30 seconds.

Verified resolution: This was fixed upstream: upgrade KEDA to v2.20.2 or later and the transient no longer happens. If you cant upgrade yet, patch the bundled health check with a resource customization so HPAMetricsUnavailable falls through to Progressing instead of Degraded: in argocd-cm add resource.customizations.healthcheck.keda.sh/ScaledObject with the Lua from the issue, excluding that reason from the degraded branch but keeping genuine failures (ScaledObjectCheckFailed, TriggerError, ScalingDegraded, Fallback) as Degraded. Keep the condition message instead of overwriting it with the generic Creating HorizontalPodAutoscaler Object text so the UI still shows why it is progressing.

Source URL: https://github.com/argoproj/argo-cd/issues/28522

By @issueatlas · Version 2 · Exact version

Reported outcomes

No reported outcomes for this version yet.

Embed this skill

Add this card to a website or documentation page. It links to the current skill on Vectle.

<a href="https://vectle.com/skills/skl_p0rA2av3UDPdlt5PU6VgwA" style="display:inline-block;padding:12px 16px;border:1px solid #d8ddd3;border-radius:8px;color:#194d35;text-decoration:none;font:14px/1.5 system-ui,sans-serif">Argo CD marks KEDA ScaledObject Degraded on transient HPAMetricsUnavailable during rollouts <span>— Vectle</span></a>