Releases: elastic/cloud-on-k8s
Releases · elastic/cloud-on-k8s
1.0.0-beta1
Breaking changes:
- Rename nodes to nodeSets, nodeCount to count #1843
- Remove setVmMaxMapCount from Elasticsearch CRD #1839 (issue: #1712)
- Bump CRD versions to v1betav1 #1782
- Add webhook check for PVC modification #1517 (issue: #1293)
- Refactor nodes orchestration to rely on StatefulSets #1463 (issue: #1299)
- Orchestrate zen1 and zen2 settings for StatefulSets #1262 (issue: #1173)
- Implement rolling upgrades with StatefulSets #1219
Enhancements:
- Log validation errors #1942
- Disable serving prometheus metrics by default #1930
- Add StatefulSets info in eck-dump tool #1920
- Rename statefulset label name with the "-name" suffix #1913
- Rename Elasticsearch Pod hash labels #1912 (issue: #1911)
- Force upgrade pods of a same StatefulSet #1888 (issue: #1847)
- Add best effort spec validation #1887
- Force upgrade all Pods if non-ready #1827 (issue: #1799)
- Set Kibana default memory requests and limits to 1Gi #1818 (issue: #1454)
- Set APM Server default memory requests and limits to 512MiB #1815 (issue: #1454)
- Add support for MaxSurge and MaxUnavailable during scaling #1812
- Set Elasticsearch Pod default memory limit to 2Gi #1810 (issue: #1454)
- Use synchronous HTTP calls to fetch shards #1778
- Bind ES Pod readiness to a node-local endpoint #1777 (issues: #916, #1748)
- Set a dynamic MinAvailable value for the default PDB #1775 (issues: #916, #1773, #1774)
- Rolling Upgrade: Support master node type change #1745
- Remove PersistentVolumeClaims when removing Elasticsearch nodes #1736 (issue: #1288)
- Ignore synced flush conflict during rolling upgrades #1733
- Upgrade to Kubebuilder v2 #1723 (issues: #1188, #1604)
- Support more secret volume fields in secure settings #1665
- Validate Elasticsearch resource names #1647
- Allow for multiple user specified secure settings secrets #1627
- Allow users to disable TLS for HTTP in the Elasticsearch spec #1623
- Preserve labels and annotations on public cert secrets #1580
- Generate events for reconcililation errors #1578
- Remove ElasticsearchInlineAuth from associations #1566
- HTTP: Add support for external CA #1538
Bug fixes:
- Manifest generation: prefix objects with YAML document separator #1901
- Set BlockOwnerDeletion to false on PVCs #1891
- Parse /_cat/shards output #1840
- Set expectations when upscaling a StatefulSet #1813 (issue: #1678)
- Fix handling of HTTP CA #1742
- When using OnDelete strategy don't rely on current/updated revision #1732
- Remove StatefulSet config secret and headless service on delete #1730 (issue: #1713)
- Update ES resource phase to operational #1719
- Always enable file based user auth #1698
- Take master change budget into account when adding new StatefulSets #1682
- Re-bootstrap single master 7.x clusters when upgrading from 6.x #1681
- Make sure there is no ongoing Pod deletion before downscaling #1534 (issue: #1523)
- Prevent HealthCheckNodePort updates #1519
- Fix version validation #1480
- Workaround controller-runtime webhook upsert bug #1337
- Add console output to standalone APM sample #1321
- Allow license secret webhook to fail #1301
- Add HTTP certs to config checksum #1267
- Respect TLSOptions for APM server #1246
0.9.0
Features:
- APM configuration #1181
- Kibana HTTP certificates #1148
- Fix status comparison in Openshift #1136
- Utility to dump info to diagnose problems #1115
- Expose Kibana keystore for secure settings #1063
- Add support for Kibana configuration #1030 (issue: #740)
- Add basic telemetry #1023
- Add support for using custom HTTP certificates in ES #1013
- Build Kibana Deployment pod template on top of user-provided podTemplate #952 (issue: #772)
- Build ES pod on top of user-provided podTemplate #943 (issue: #765)
Enhancements:
- Relax Elasticsearch validation hook failure policy #1386
- ES samples improvements #1375
- Ignore resources reconciled by older controllers #1286 (issue: #1241)
- Remove process manager #1249
- Change CN and SAN naming #1243
- Remove APM server defaults and allow standalone operation #1231
- Warn if unsupported distribution #1228
- Add operator version to resources #1224 (issue: #1137)
- Move the ES keystore update into an init container #1202
- Set a default memory resources request for Elasticsearch #1180 (issue: #1097)
- Add flag for debug logs #1166 (issue: #305)
- Remove RemoteCluster and TrustRelationship #1164
- Update operator information #1139
- Allow users to specify transport certificate authorities #1133
- Improve pod-pvc labelling and comparison #1129
- Improve Kibana deployment comparison with a hash #1110
- Improve ES pods comparison using a hash of the spec #1103 (issues: #851, #1046)
- Move prepare-fs script to scripts configmap #1070
- Change default mode to allow running as non root #1069
- Remove plugins from prepare-fs script in init container #1060
- Move readiness probe script to configmap #1054
- Update base operator Docker image to centos:7 #1043
- Add rbac on */finalizers to support Openshift #1041
- Do not require root privileges when not needed #1039
- Remove JVM DNS cache policy tweaks #1033 (issues: #323, #931)
- Use persistent storage by default for ES data #1028 (issue: #913)
- Provide a default Affinity for ES Pods. #1026 (issue: #915)
- Create a default PDB for ES Pods. #1025 (issue: #914)
- Normalize Elasticsearch pod related naming #1024 (issue: #547)
- Provide default images and volume mounts to user-provided init containers #1022
- Remove enterprise license CRD #1018
- Allow additional labels to be set on the tranport cert secrets #1008
- Don't override user-provided labels from Kibana podTemplate #1006
- Make .spec.http.service fully configurable by default #1004
- Remove default discovery service #995
- Move Kibana configuration in a Secret #977
- Set a default subdomain=es.name and name=pod.name on new Pods #974
- Remove cluster license CRD #969
- Don't override user-provided labels #957
- Remove User CRD #948
- Separate HTTP/Transport certificates and CAs #929
Bug fixes:
- Parameterize keystore settings addition #1401
- Update the way to add settings in the ES keystore #1377
- Fix enable-debug-logs flag in the operator manifest #1372 (issue: #1371)
- Workaround controller-runtime webhook upsert bug #1337
- Don't rely on buggy metaObject Kind #1324
- Add console output to standalone APM sample #1321
- Allow license secret webhook to fail #1301
- Secure settings finalizer: do not rely on object.GetObjectKind() #1279
- Do not set initial_master_nodes if cluster has been bootstrapped #1272
- Add http certs to config checksum #1267
- Respect TLSOptions for apm server #1246
- Normalize the raw config when creating canonical configs #1208
- Update PodSpec with Hostname from PVC when re-using #1204
- Error message typo #1179
- Do not move files when preparing the fs, copy them instead #1128
- Set the operator-namespace to default when running locally #1125
- Respect volume name when reusing PVCs #1122
- Fixes issue where explicit non-pvc volume in template was ignored #1079
- Fix scripts invocation #1077
- Support all 7.x versions of Elasticsearch #1075
- Standard might not be the default storage class #1061
- PVC may be stolen by an other pod #1052
- Adds missing Namespace to some client.ListOptions structs #1020
- Add missing namespace to Pod Listing when reconciling transport secrets #1019
- Consider more defaulted values when comparing Services for reconciliation #1011
- Fix object deletion in association #1009
- Move directories only if not empty in prepare-fs script #937
