Upgrade Guide - Scylla 1.0 to 1.1 for Ubuntu

This document is a step by step procedure for upgrading from Scylla 1.0 to Scylla 1.1, and rollback to 1.0 if required.

Applicable versions

This guide covers upgrading Scylla from the following versions: 1.0.x to Scylla version 1.1.y on the following platform:

  • Ubuntu 14.04

Upgrade Procedure

A Scylla upgrade is a rolling procedure which does not require full cluster shutdown. For each of the nodes in the cluster, serially (i.e. one at a time), you will:

  • backup the data
  • check your current release
  • backup configuration file
  • download and install new Scylla packages
  • gracefully restart Scylla
  • validate that the upgrade was successful

Apply the following procedure serially on each node. Do not move to the next node before validating the node is up and running with the new version.

During the rolling upgrade it is highly recommended:

  • Not to use new 1.1 features
  • Not to run administration functions, like repairs, refresh, rebuild or add or remove nodes
  • Not to apply schema changes

Backup the data

Before any major procedure, like an upgrade, it is recommended to backup all the data to an external device. In Scylla, backup is done using the nodetool snapshot command. For each node in the cluster, run the following command:

nodetool snapshot

Take note of the directory name that nodetool gives you, and copy all the directories having this name under /var/lib/scylla to a backup device.

Upgrade steps

Backup configuration file

sudo cp -a /etc/scylla/scylla.yaml /etc/scylla/scylla.yaml.backup-1.0
sudo rm -rf /etc/apt/sources.list.d/scylla.list

Download and install the new release

Before upgrading, check what version you are running now using dpkg -s scylla-server. You should use the same version in case you want to rollback the upgrade. If you are not running a 1.0.x version, stop right here! This guide only covers 1.0.x to 1.1.y upgrades.

To upgrade:

  1. Update the Scylla deb repo to 1.1
  2. Install
sudo apt-get update
sudo apt-get upgrade  scylla-server scylla-jmx scylla-tools

Answer ‘y’ to the first two questions and ‘n’ when asked to overwrite scylla.yaml.

Gracefully restart the node

nodetool drain
sudo service scylla-server restart

Validate

  1. Check cluster status with nodetool status and make sure all nodes, including the one you just upgraded, are in UN status.
  2. Check /var/log/upstart/scylla-server.log and /var/log/syslog to validate there are no errors.
  3. Check again after 2 minutes, to validate no new issues are introduced.

Once you are sure the node upgrade is successful, move to the next node in the cluster.

Rollback Procedure

The following procedure describes a rollback from Scylla release 1.1.x to 1.0.y. Apply this procedure if an upgrade from 1.0 to 1.1 failed before completing on all nodes. Use this procedure only for nodes you upgraded to 1.1

Scylla rollback is a rolling procedure which does not require full cluster shutdown. For each of the nodes rollback to 1.0, you will:

  • retrieve the old Scylla packages
  • drain the node
  • restore the configuration file
  • restart Scylla
  • validate the rollback success

Apply the following procedure serially on each node. Do not move to the next node before validating the node is up and running with the new version.

Rollback steps

download and install the old release

  1. Remove the old repo file.
sudo rm -rf /etc/apt/sources.list.d/scylla.list
  1. Update the Scylla deb repo to 1.0
  2. install
sudo apt-get update
sudo apt-get remove --assume-yes scylla-server scylla-jmx scylla-tools
sudo apt-get install scylla-server scylla-jmx scylla-tools

Answer ‘y’ to the first two questions and ‘n’ when asked to overwrite scylla.yaml.

Gracefully shutdown Scylla

nodetool drain
sudo systemctl stop scylla-server.service

Restore the configuration file

sudo rm -rf /etc/scylla/scylla.yaml
sudo cp -a /etc/scylla/scylla.yaml.backup-1.0 /etc/scylla/scylla.yaml

Start the node

sudo service scylla-server start

Validate

Check upgrade instruction above for validation. Once you are sure the node rollback is successful, move to the next node in the cluster.