57Off-peak, the autoscaler removes nodes and our Spark driver got evicted mid-run. How does scale-down choose its victims?▼mediumNewAmazon & AWSLinkedInAtlassian◆ premiumScale-downers mark a node removable only when every pod on it can leave safely. Your batch workload died because nothing marked it special. The mechanism is a filter cascade, and the fix is priorities, PDBs and annotations.Open full answer →
65The cluster adds four nodes and removes them again every ten minutes, all night, with flat traffic. Stop it.▼hardNewAmazon & AWSDatabricksDatadog◆ premiumTwo controllers with different opinions and no shared hysteresis. Flat traffic with an oscillating node count means the loop is driven by something other than demand, and the candidates are few.Open full answer →
