Announcing Scylla Manager 1.0 – Centralized Administration and Task Automation
Scylla Manager adds centralized cluster administration and recurrent tasks automation to Scylla Enterprise. Scylla Manager 1.0 includes automation of periodic repair, with future releases providing more capabilities, such as rolling upgrades, recurrent backup, and more. With time, Scylla Manager will become the focal point of Scylla Enterprise cluster management, including a GUI frontend. Scylla Manager is available for all Scylla Enterprise customers, or for a 30-day trial.
Scylla Manager includes three main parts: A Go based server, a CLI tool, sctool, to interact with it, and a highly available backend to persist its state. The server securely interacts with one or more Scylla clusters, learns the cluster topology, and runs cluster-wide tasks in a controlled and predictable way. A user can use sctool to initiate and schedule management operations, monitor their progress, or suspend them.
To provide high-availability, Scylla Manager is designed to be stateless and uses Scylla for data persistence. By default, Scylla Manager installs one local instance of Scylla as backend. For a highly available management solution, we recommend using a multi-node Scylla cluster as a backend. Although possible, we do not recommend using the managed Scylla cluster to serve as a Scylla Manager backend.
Recurrent Repairs Automation
The first production-ready release of Scylla Manager includes automation for recurrent repairs. It is highly recommended to run weekly Repair tasks on production Scylla clusters. However, until now, this task had challenges, including:
- No centralized element to manage repairs across all nodes because nodetool repair is a local operation and it was not clear if one should run repair node by node, Keyspace by Keyspace, etc
- No out-of-the-box scheduler for repairs, so each user build their own solution, not always in a highly-available manner
- No easy way to suspend a repair process and resume it later
Scylla Manager 1.0 solves all of the above problems. It comes with a built-in scheduler and executor. Each repair task is automatically broken into small chunks so that repair can be stopped and resumed at any point. And, using sctool, you can easily monitor the progress of the repair task at each point.
Once provided with a subset of nodes per cluster, Scylla Manager automatically learns the cluster topology and keyspaces, and once per week by default it creates a full cluster repair. You can view and update the scheduling using sctool. Note that one instance of Scylla Manager can manage multiple Scylla clusters.
Download Scylla Manager – for customers only, or for limited evaluation.
Get started with Scylla Manager
Using sctool, the Scylla Manager CLI tool.
Report issues with Scylla Manager using the Scylla Enterprise support channels.