v0.2.0

New Features

• Multi datacenter support. Full docs can be found at http://mesosphere.github.io/cassandra-mesos/docs/multi-dc.html
• Improved health check logic
• Fixed executor lost handling
• Fixed exception when repairing cluster
• DCOS Compatibility improvements

Updating

Users who were running Cassandra on Mesos v0.1.0 should be able to update to v0.2.0 with minimal impact. The version of Cassandra stayed at 2.1.4.

Steps

1. Shutdown scheduler running v0.1.0
2. Start scheduler running v0.2.0
3. Wait for new scheduler to come up and register with mesos
4. Manually kill each executor in the cluster one at a time
   • If Cassandra on Mesos is configured to write it's data to the sandbox (the default behavior) this will result in potential data loss
5. Cassandra on Mesos should handle the event and start new executors on each node automatically

Artifacts

marathon.json: https://downloads.mesosphere.io/cassandra-mesos/artifacts/0.2.0-1/marathon.json
tar: https://downloads.mesosphere.io/cassandra-mesos/artifacts/0.2.0-1/cassandra-mesos-0.2.0-1.tar.gz

Fixed Issues

• #114 Allow configuration of mesos user and advertised host for http server
• #115 Allow configuration of default rack and dc for GossipingPropertyFileSnitch
• #116 Add overview to Table of Contents
• #112 UninitializedMessageException when repairing cluster
• #119 Update code to ensure that currently running job is used as base of building new job state.
• #118 Update health check logic to be tolerant of nodes that have been terminated
• #120 Add functionality to sync seeds between clusters
• #123 Clarify purpose of env variables for configuration
• #125 Update LD_LIBRARY_PATH of executors running on DCOS
• #124 Accept offers for default role * in addition to the specified mesosRole
• #121 Cassandra tasks getting LOST status in mesos
• #129 Update executor lost handling to properly clear executor information
• #130 Add docs for multi-dc support

Contributors

Thanks goes to the following individuals (in alphabetic order)

• @BenWhitehead
• @dmitrypekar
• @flosell
• @gabrielhartmann
• @joel-hamill

v0.1.0...v0.2.0

v0.1.0

This is the first release of the Apache Mesos framework for Apache Cassandra.

A high level list of features that have been implemented

• Scheduler will bootstrap a keep a ring running
• Restart a node that has crashed
• Uses default Cassandra ports from Mesos Resource Offers
• Executor monitors health of a running node
• Scheduler provides all configuration to nodes
• Support adding nodes to the ring
• Able to be ran via Marathon, all builds output a marathon.json
• Expose REST API to manage Cassandra Cluster
• Expose cluster health check via Marathon Health check
• Expose api resource to view cluster state
• Expose api resource to view scheduled tasks
• Expose api resource to view running periodic tasks
• Expose api resource to view cluster configuration
• Expose api resource to manage nodes and seeds
• Expose api resource to manage running cleanup and repair
• Self-contained, no dependency on internet resources after scheduler has started
• Mesosphere DCOS Cassandra Package

Full Documentation: http://mesosphere.github.io/cassandra-mesos/

Artifacts

marathon.json: https://downloads.mesosphere.io/cassandra-mesos/artifacts/0.1.0-1/marathon.json
tar: https://downloads.mesosphere.io/cassandra-mesos/artifacts/0.1.0-1/cassandra-mesos-0.1.0-1.tar.gz

Thanks

A special thanks goes to @snazy, @ssk2 and @BenWhitehead for their contributions to this release.