LavaFlow creates useful reports on the usage of high-performance computing clusters. It takes data from the batch scheduling system, monitoring, and other tooling, and creates reports which help administrators, managers, and end users better understand their cluster environment. The reports are modular, and new modules are easy to create using templates and Django's query set API. LavaFlow uses human-readable RESTful URLs, making it easy to automate and share links to reports.
JASocket is a lock-free, scalable, and robust server framework with no single point of failure. Servers are run on a cluster of nodes. Servers interact with other servers using mobile agents, which reduces the number of messages and thus reduces the overall system latency. Administration is handled via ssh.
JAConfig implements an eventually consistent distributed key/value database for managing a JASocket cluster. Also included are Quarum for tracking when a quorum of hosts is present, Ranker for determining which nodes are least loaded, ClusterManager for starting up other servers, and Kingmaker, which decides which node is to run ClusterManager. JAConfig is lock-free, actor-based, and has no single point of failure.
Strategico is an engine for running statistical analysis over groups of time series. It can manage one or more groups (projects) of time series: by default, you can get data from a database or CSV files, normalize them, and then save them inside the engine. The first statistical analysis implemented inside Strategico is the "Long Term Prediction": it automatically finds the best model that fits each time series. Some of the models implemented are mean, trend, linear, exponential smoothing, and Arima. Strategico is scalable: the statistical analysis over each time series (of a project) can be run separately and independently. It is suggested that you set up an HPC Cluster (High Performance Computing) and/or use a resource scheduler like slurm. It is developed with R, one of the most famous statistical languages.
OpenSVC is a 'service' manager, as in clustered service manager. Services are described as collections of resources (IP, disk groups, filesystems, file synchronizations, and application launchers). Services can be started, stopped and queried for status, providing a consistent command set for wildly different service integrations. Services can be administered using a stand-alone free software stack deployed on the nodes (nodeware). Service configurations, status, and logs are pushed to a central database coupled to a Web front-end (collector).
Libfairydust is a small wrapper library intended for use with GPU clusters that 'hijacks' CUDA and OpenCL calls. It can be used to 're-route' calls to a certain GPU, so a process requesting GPU#0 might end up running on GPU#4 without knowing (or caring) about it. This works completely transparently and does not need any sort of 'cooperation' from the application, changes to code, or relinking.
Gossimon is a gossip-based distributed monitoring system for a cluster of Linux nodes. Each node in the cluster periodically send information about itself and others to a randomly selected node. This way, each node constantly receives information about cluster nodes. This information is locally maintained (constantly updated) by each node and can be used by various clients for monitoring and resource allocation. The gossip protocols used by gossimon are very robust to node failure, and the information quality is hardly degraded even when large parts of the cluster are taken down. The package contains: infod (the daemon responsible for collecting and sending information for other nodes); mmon (a curses-based monitoring client displaying information about cluster nodes); and infod-client (a command-line client that retrieves cluster information in XML format).
Lingua Server is a fast multi-protocol application service server in Java. Its main features are very fast response times, a pluggable protocol and transport architecture, very easy application service development (MVC), built-in cluster support using multicast, auto-discovery, and distributed shared objects.