All notable changes to this project will be documented in this file.
- switchover on a runner node may fail if runner fiber was dead
- task running on several workers if ttr exceeded due to long execution
- older tasks are not displayed in the UI
- completed tasks not being compacted if there’s an old pending task
- ability to launch massive amounts of tasks via task.start()
- task storage statistics for monitoring purposes
- broken storage upgrade from task v0.4.0
- Performance improvements.
- Fixed performance degradation on put operation when task storages are ‘full’.
- Removed second app title in UI.
- Added arguments column in task log table.
- Added ability to launch anonymous tasks a. k. a. jobs
- Added ability to specify retry strategies for tasks
- Added ability to specify defaults for runner/storage options via task.init
- Bugfixes and minor improvements
- Ability to specify task args in config
- several bugfixes
- task storages can now be set up on vshard storages
- default task configuration can be set up in task.init()
- task now depends on cartridge instead of cluster
- cartridge dependency bumped to 1.0.0
Added: - Cluster-aware task storage role - Cluster-aware task runner role
Changed: - Module initialization API changed: now task scheduler role does not start runner and storage by default; you must handle it via role dependencies or via respective cluster roles
Added: - Basic functionality - scheduler, local task storage, local task runner, cluster role, cluster UI