Modern cluster managers orchestrate large-scale services and resources through a set of controllers, each managing a specific part of the cluster by iteratively reconciling the cluster states into the desired states. However, controllers are prone to various state inconsistencies stemming from asynchrony, concurrency, and failures, posing significant challenges in reliable cluster operation. Our analysis of 51 consistency bugs in Kubernetes controllers reveals that many issues remain unresolved, and even proposed fixes are often rejected for backward incompatibility or performance loss.

Hyper-parameter optimization is crucial for pushing the accuracy of a deep learning model to its limits. However, a hyper-parameter optimization job, referred to as a study, involves numerous trials of training a model using different training knobs, and therefore is very computation-heavy, typically taking hours and days to finish. We observe that trials issued from hyper-parameter optimization algorithms often share common hyper-parameter sequence prefixes. Based on this observation, we propose Hippo, a hyper-parameter optimization system that reuses computation across trials to reduce the overall amount of computation significantly. Instead of treating each trial independently as in existing hyper-parameter optimization systems, Hippo breaks down the hyper-parameter sequences into stages and merges common stages to form a tree of stages (a stage tree). Hippo maintains an internal data structure, search plan, to manage the current status and history of a study, and employs a critical path based scheduler to minimize the overall study completion time. Hippo applies to not only single studies but multi-study scenarios as well. Evaluations show that Hippo's stage-based execution strategy outperforms trial-based methods for several models and hyper-parameter optimization algorithms, reducing end-to-end training time by up to 2.76X (3.53x) and GPU-hours by up to 4.81X (6.77x), for single (multiple) studies.

Conversational robots that exhibit human-level abilities in physical and verbal conversation are widely used in human-robot interaction studies, along with the Wizard of Oz protocol. However, even with the protocol, manipulating the robot to move and talk is cognitively demanding. A preliminary study with a humanoid was conducted to observe difficulties wizards experienced in each of four subtasks: attention, decision, execution, and reflection. Apprentice of Oz is a human-in-the-loop Wizard of Oz system designed to reduce the wizard's cognitive load in each subtask. Each task is co-performed by the wizard and the system. This paper describes the system design from the view of each subtask.