Kubernetes operator that provides control plane for managing Apache Flink applications
Full Changelog: https://github.com/lyft/flinkk8soperator/compare/v0.5.0...v0.6.0
This release includes a number of feature additions and a couple of bug fixes. With this release, the operator now supports zero-downtime deployments by enabling a BlueGreen
deployment mode. This is an opt-in feature and the modified state machine for BlueGreen deployments can be found here. Other feature enhancements include a SavepointDisabled
mode to opt-in to updates without savepoints and the ability to set pod tolerations and security context in the CRD.
Thanks to Kelly Smith, Anand Swaminathan, Micah Wylde, Ravi Magham, Lakshmi Rao and Nasrudin Bin Salim for contributing to this release.
This release includes a number of bug fixes and backward incompatible changes. Includes change to enable the status subresource for the FlinkApplication CRD and way SavepointInfo
is updated by the operator.
Please make sure to update the CRD and this update should not be deployed to a cluster where there are active flinkapplication updates occurring — i.e., all flinkapplications should be in a Running or DeployFailed state.
Thanks to Micah Wylde, Lakshmi Rao, Thomas Weise, Chaoran Yu, Temo Ojeda, Marcelo Da Cruz Pinto and Prem Santosh for contributing to this release.
Changes in this release:
This release includes a number of bug fixes. It also includes a breaking change in how Flink heap configurations are rendered. Previously this config was rendered incorrectly, which caused Flink to ignore it. This change will also cause all Flink applications to be redeployed when the operator is updated.
Thanks to Anand Swaminathan, Chaoran Yu, Lakshmi Rao, Marcelo Da Cruz Pinto, and Jonathan Miles for contributing to this release.
Changes in this release:
This releases bumps the flink application version to BETA - bug fixes from Alpha, added features.
373cb21 [Backward Incompatible] Add env variables for Flink managers and fix for JM HA (#45) 9053e4f Fix unit test from merge to master bb7c8dd Enable allowNonRestoredState in flink CRD (#43) 1a62f03 [STRMCMP-558] Event improvements (#44) 8d847d7 Improve staleness logic by improving error handling and retrying a subset of errors (#32) 4297702 Find the correct deployments for readiness and status checks (#42) 7fda241 Add more metrics to Flinkk8soperator (#41) 9a95850 Initialize Klog explicitly using InitFlags (#40) f920664 [Backward Incompatible] Upgrade controller runtime and Multi Namespace support for Flink operator (#39) 9aaa992 Add support for non restored state on job submission (#38) cde4f38 Update logic around number and healthy task managers in status (#37) 6d92339 Use EventRecorder instead of emitting events manually (#36)
6dd5f55 Adding openAPIV3Schema Validation for the Flink application custom resource (#33) 2f66087 [STRMCMP-544] Allow FlinkApplications to be deleted when job submission fails (#30) 1456182 [STRMCMP-509] Make old cluster clean-up idempotent (#27) 93f0e01 Fix Resty logging, timeout and retry configuration (#28) 9e35f64 Made sample app run on OpenShift (#26) 2e53557 Added doc on mounting volumes (#25) d9dd430 Handle invalid phases in the Flink state machine (#23) 1616e73 Add node selectors to job and task config (#7) 8bfdcff Upgrade to latest version of FlyteStdlib (#21) b8ff17d Minor fixes and improvements (#17) d1d1176 Log operator config on startup (#11) 75f04e5 Add update/delete verb for service (#8) 5dbd1bd Added finalizer permissions (#12) d3ff566 Add additional printer columns (#16) 390fa96 Fix the travis condition to make docker hub push on merge to master (#15) f3c05db Add Lakshmi as codeowner (#14) 5aa8a13 Optimize image pushes and integration tests (#5)
Initial release of the flink operator