The Linux default scheduler's fairness-oriented design frequently preempts short-running serverless functions, degrading Function-as-a-Service (FaaS) performance and increasing user costs. Existing solutions to remedy this problem propose custom OS scheduling policies based on custom kernels, external frameworks, or user-space polling. To address these limitations, we explore sched_ext, a recent Linux kernel feature providing an extensible scheduler framework via the Extended Berkeley Packet Filter (eBPF). We evaluate 11 prebuilt sched_ext schedulers under production-like FaaS workloads and show that 9 out of 11 policies outperform the default Linux scheduler (EEVDF) in both performance and user costs, reducing turnaround time by up to 61.1% and user costs by up to 51.3%. In this work, we demonstrate the significant impact of OS scheduling on FaaS environments and identify successful mechanisms to guide the design of future specialized schedulers.
Saman Akbari, Manfred Hauswirth· Proceedings of the 17th ACM...· 0 citations
LLMs remain unreliable for long-horizon planning, often generating logically inconsistent or non-applicable plans. Recent hybrid methods instead translate natural language into the Planning Domain Definition Language (PDDL), allowing symbolic planners to produce verifiable plans. However, existing methods frequently rely on rigid generation pipelines, a partial PDDL definition, or human feedback. Furthermore, their evaluation is hindered by the lack of standardized benchmarks with automated verification. To address these limitations, we present PDDLCoder, an agentic framework for PDDL generation from natural language that iteratively generates, analyzes, and refines planning specifications. We further introduce NL-pddlgym, a benchmark dataset comprising 711 planning problems across 23 domains with executable gym environments for the automated verification of plan applicability. Experiments on the NL-pddlgym test set containing 106 problems across 4 held-out domains show that PDDLCoder generates applicable plans for 89.6\% of tested planning problems. This improves upon our adaptations of previous PDDL generation methods, which achieved up to 45.3\%, and outperforms direct LLM planning approaches, which reached up to 74.5\% on the same test set. Our work demonstrates the effectiveness of agentic PDDL generation for planning and establishes a reproducible benchmark for future research on LLM-assisted symbolic planning.
Veit Laule, Jiangtao Shuai, Manfred Hauswirth et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
