An AWS engineer reported a significant drop in PostgreSQL throughput on Linux 7.0, with performance reduced to approximately half of its previous capability. Benchmark tests showed that the removal of the PREEMPT_NONE scheduling option was the main cause of this regression. On a 96-vCPU Graviton4 instance, throughput measured at just 0.51x compared to earlier kernel versions. Salvatore Dipietro from Amazon/AWS conducted benchmarking analysis of PostgreSQL 17, revealing that Linux 7.0 delivered only 0.51x the throughput of its predecessors. The root cause was traced to kernel commit 7dadeaa6e851, which eliminated PREEMPT_NONE as the default option, leading to increased contention due to the new PREEMPT_LAZY model. Profiling data indicated that 55% of CPU time is consumed by spinning in PostgreSQL’s spinlock, causing significant performance degradation. When a revert patch was applied, throughput rebounded to 1.94x the baseline. The decision to restrict preemption modes in Linux 7.0 aimed to address issues within the kernel's scheduling model. Dipietro proposed a patch to restore PREEMPT_NONE, but kernel developers suggested PostgreSQL adopt the rseq time slice extension instead. Database operators running PostgreSQL on Linux face potential performance reductions with the upgrade to Linux 7.0.