Enhanced automatic scaling for cluster components

We have rolled out improvements to our platform’s automatic scaling capabilities. These optimizations allow cluster components to better adapt to the varying workloads and demands of our customers’ environments without the need for manual adjustments.

What Changed

Expanded VPA Scaling Range

The Vertical Pod Autoscaler (VPA) can now scale components within a broader range. We’ve increased the maximum limit from 2x to 4x the base resource settings, giving the scaler more headroom to respond to resource-intensive environments and OOM events without manual intervention.

VPA Applied to Additional Components

We’ve reviewed all platform components and ensured VPA is consistently applied where appropriate. Components that benefit from vertical scaling now have proper VPA configurations, either through their Helm charts or custom definitions.

Improved Loki Defaults

To better accommodate customer logging needs, we’ve increased the default limits for Loki ingestion, including ingestion_rate_mb, ingestion_burst_size_mb, per_stream_rate_limit, and per_stream_rate_limit_burst.

Why This Matters

These changes reduce the likelihood of resource-related alerts and service degradation during traffic spikes or increased cluster utilization.

No action is required on your part—these improvements are being rolled out automatically to all clusters.