All notable changes to this project will be documented in this file. This project adheres to Semantic Versioning.
- Do not change ES binary group
- Consistent filenames for releases
- Null check resources
- Extra information in DiscoveryInfo packet
- Framework uses official Elasticsearch image or binaries. Users are able to upgrade ES without framework upgrade.
- CLI: Executor image names have been renamed to Elasticsearch image names.
- Executors will quit when scheduler is killed with SIGTERM.
- Upgrade minimesos to 0.6.1
- Add option to specify exact ES image
- External Volumes support
- Remove executor. Use official binaries
- Use official binaries
- Shut down executors when scheduler receives SIGTERM
- Upgrade and refactor dependencies
- Links to published jars
- Node specific data directory for shared network drives
- Users are now required to have a path on the end of the zk url. Previously, they were not.
- Stale data when scaling nodes
- Health check issue
- Framework hard codes the ZK endpoint /mesos
- With large amounts of data, when scaling, start up times out
- elasticsearch-executor name
- In docker mode, user has to place yml file in /tmp/config
- Downloading yml settings file from https source
- Update to minimesos 0.5.0+
- Upgrade to elasticsearch 2.1.1
- Allow a user to specify his own http/transport ports
- Manual horizontal scaling
- Networking overhaul. Now uses standard ES zen for clustering.
- shouldNotLoseDataWhenScalingDown system test occasionally fails
- Minimesos hostnames do not point to running containers
- Fix Jenkins system test failures
- Option to use IP addresses instead of hostname
- Documentation
- Jitpack cannot build jar beacuse of docker build
- Remove dependency on ES Zookeeper
- Sanity check for framework registration before getting cluster state
- Migrate ES system tests to new MiniMesos API
- System test hardening and refactoring (various)
- Cleaned debug logging (various)
- Executor should report its own IP address to scheduler refactoring
- TASK_KILLED was not considered an error state, so executors would not be removed
- GUI configuration NPE due to new jar options
- Remove 'mesos/elasticsearch-base' from docker push on release
- Implement Mesos Authorisation
- Allow users to run without docker
- Shrink docker images
- Replace JVM with OpenJDK runtime
- Fix readthedocs documentation links to github bug documentation
- Refactor scheduler to further testing refactoring
- Scheduler Search API endpoint returns "java.util.NoSuchElementException" API bug
-
Healthcheck thread not stopped during Unit tests bug testing
-
When executors are lost, the status is not propagated back to zookeeper bug
- GUI correctly represents state of cluster
- Request resources as role
- Zookeeper unable to create /es node - Incorrect zk formatter
- Upgraded to Mini-Mesos 0.2.6 with fast-fail logging
- Full configuration of framework via CLI
- User can override ES settings with their own elasticsearch.yml
- Data is now stored in Docker volumes
- Support multiple Elasticsearch clusters on single Mesos cluster
- Configurable scheduler and executor images
- Support separate Zookeeper cluster for the framework
- Fixed UI redirection bug
- Removed zk, m, n and ram parameters from old CLI
- CLI code now uses JCommander library