Abstract: Aspects of data management are described. During an update procedure for serially updating a cluster of storage nodes, a storage node of the cluster of storage nodes may enter a quiescent state. While in the quiescent state, the storage node may refrain from obtaining new jobs and may continue to execute jobs that were initiated at the storage node prior to entering the quiescent state. The storage node may enter the quiescent state while another storage node enters an update state for installing the update version. The storage node may also post, to a job queue, jobs running at the storage node that are terminated at an end of the quiescent state.

Abstract: A database management system may schedule, according to a first job allocation configuration, a first job allocation for a first computing node, where the first computing node is scheduled to perform a parent job. In some examples, the first job allocation configuration prevents scheduling the parent job on a computing node running a newer code version than a set of computing nodes running the set of child jobs until the set of child jobs terminates. The database management system may further schedule, according to a second job allocation configuration, a second job allocation for a second computing node, where the second computing node is scheduled to perform a child job of the set of child jobs. In some examples, the second job allocation configuration prevents scheduling the child job on computing nodes having a newer code version than the first computing node performing the parent job.

Abstract: Methods and systems for improving the performance of a distributed job scheduler by dynamically splitting and distributing the work of a single job into parallelizable tasks that are executed among multiple nodes in a cluster are described. The distributed job scheduler may split a job into a plurality of tasks and assign the tasks to nodes within the cluster based on a time remaining to complete the job, an estimated time to complete the job, and a number of identified healthy nodes within the cluster. The distributed job scheduler may monitor job progress over time and adjust (e.g., increase) the number of nodes used to execute the plurality of tasks if the time remaining to complete the job falls below a threshold amount of time or if the time remaining to complete the job minus the estimated time to complete the job falls below the threshold amount of time.

Abstract: Methods and systems for improving the performance of a distributed job scheduler by dynamically splitting and distributing the work of a single job into parallelizable tasks that are executed among multiple nodes in a cluster are described. The distributed job scheduler may split a job into a plurality of tasks and assign the tasks to nodes within the cluster based on a time remaining to complete the job, an estimated time to complete the job, and a number of identified healthy nodes within the cluster. The distributed job scheduler may monitor job progress over time and adjust (e.g., increase) the number of nodes used to execute the plurality of tasks if the time remaining to complete the job falls below a threshold amount of time or if the time remaining to complete the job minus the estimated time to complete the job falls below the threshold amount of time.