Release notes for 1.12.0

Container images

docker.io/scylladb/scylla-operator:1.12.0

Changes By Kind (since 1.11.3)

Feature

• Update ScyllaDB Monitoring to 4.4.5 (#1456,@tnozicka)
• Update dependencies (#1480,@tnozicka)
• Update dependencies (#1494,@tnozicka)
• Make E2E suite expose ScyllaClusters on PodIPs by default (#1508,@rzetelskik)
• Bump gocql to scylladb/gocql@v1.12.0 (#1528,@rzetelskik)
• Collect CRDs in must-gather (#1562,@tnozicka)
• Automatically create explicit dest-dir for must-gather (#1563,@tnozicka)
• Update dependencies (#1581,@tnozicka)
• Gracefully evict Pods on ScyllaDB node replacements (#1598,@tnozicka)
• Update grafana UBI image (#1602,@tnozicka)
• Update Scylla Manager to 3.2.5 (#1603,@tnozicka)
• Update dependencies (#1604,@tnozicka)
• Update ScyllaDB to 5.4.0 (#1606,@tnozicka)
• Update ScyllaDB version to 5.2.11 (#1610,@tnozicka)
• Automatically detect number of cpus from cgroup in Scylla Operator binaries (#1612,@zimnx)
• Support scyllaArgs in Helm chart (#1613,@alexgenco)
• Make haproxy manifests work on both amd64 and arm64 (#1654,@tnozicka)
• Update local volume provisioner to v0.3.0 (#1656,@tnozicka)
• Bump github.com/go-git/go-git/v5 from 5.10.1 to 5.11.0 (#1662,@dependabot)
• Inherit and customize ScyllaCluster labels and annotations (#1664,@tnozicka)
• Update dependencies (#1677,@tnozicka)
• Extend NodeConfig with loop device management (#1691,@zimnx)
• Switch from in-tree mermaid client to out-of-tree manager client (#1702,@rzetelskik)
• Drain the node before Pod is terminated (#1707,@zimnx)
• Collect namespace where NodeConfig Pods are executed (#1715,@zimnx)
• Fully support Alternator API and deprecate the old port option (#1719,@tnozicka)
• Get cpuset from kubelet PodResources API (#1724,@zimnx)
• Bump default ScyllaDB utils image to 5.4.2 (#1725,@zimnx)
• Add identity service to ScyllaDB serving certs (#1728,@tnozicka)
• Add server-side printers to our APIs (#1730,@tnozicka)
• Enable internode compression by default (#1734,@zimnx)
• Extend ScyllaCluster status with aggregated data for all racks (#1737,@tnozicka)
• Add minReadySeconds and podReadinessGates to ScyllaCluster spec (#1738,@zimnx)
• Update dependencies (#1757,@tnozicka)
• Change default minReadySeconds to 5 seconds (#1781,@zimnx)
• Update dependencies (#1806,@tnozicka)
• Rollback Scylla utils image version to 5.4.0 (#1829,@zimnx)

Bug

• Evaluate special 'now' terms in Scylla Manager tasks only on task creation (#841,@zimnx)
• Explicitly set TLS CQL shard aware port (#1450,@zimnx)
• Fix namespaced must-gather (#1453,@tnozicka)
• Set ScyllaCluster placement requirements on cleanup Jobs (#1493,@zimnx)
• Fix error aggregation in scyllacluster controller job sync (#1569,@rzetelskik)
• Adjust image references for introspection with containerd (#1619,@tnozicka)
• Fix StatefulSet rolled out condition (#1624,@zimnx)
• Rollback StatefulSet selector (#1639,@zimnx)
• Determine host for Scylla client using ScyllaCluster expose options (#1641,@rzetelskik)
• Loosen validation for LoadBalancerClass in NodeServiceTemplate (#1649,@zimnx)
• Support update of StatefulSet selector (#1655,@zimnx)
• Revert introspection (#1685,@tnozicka)
• Replace nowSafety buffer with zero-value datetime in manager task scheduling (#1706,@rzetelskik)
• Fix metadata projection of StatefulSet PVC template (#1712,@zimnx)
• Collect only ServiceMonitors managed by given ScyllaDBMonitoring (#1717,@zimnx)
• Fix syntax of ScyllaCluster PreStopHook (#1727,@zimnx)
• Change minTerminationGracePeriodSeconds to *int32 (#1740,@zimnx)
• Stop Scylla Manager task before deletion (#1744,@zimnx)
• Always use default minReadySeconds so identity service works (#1746,@tnozicka)
• Return empty list of loop devices when losetup doesn't print anything (#1782,@zimnx)
• Fix alternator validation 1.12 (#1792,@tnozicka)
• Pass auth token on Scylla Manager cluster update (#1803,@rzetelskik)
• Explicitly disable CQL over TLS in Scylla Manager cluster integration (#1807,@rzetelskik)
• Fix update action loop break in manager controller (#1822,@rzetelskik)
• Decouple Alternator certificates from CQL TLS feature gate (#1842,@tnozicka)

Documentation

• Enable building v1.11 documentation (#1445,@zimnx)
• Add Multi Datacenter ScyllaDB cluster deployment documentation (#1466,@rzetelskik)
• Fix example file in 'Setting up nodes for ScyllaDB' of 'Deploying Scylla on EKS' documentation page (#1496,@rzetelskik)
• Fix steps ordering in EKS and GKE documentation pages (#1497,@rzetelskik)
• Add support and must-gather documentation (#1504,@tnozicka)
• Fix preview target in docs readme (#1507,@tnozicka)
• Unify docs index pages and make sure they are always in sync with the menu (#1509,@tnozicka)
• Describe Manager and Monitoring in MultiDC docs (#1513,@zimnx)
• Avoid service enpoints size limit in monitoring examples (#1515,@tnozicka)
• Make example ScyllaCluster in EKS docs consistently use scylladb-local-xfs storage class (#1527,@rzetelskik)
• Remove myst.xref_missing suppression in Sphinx config (#1529,@rzetelskik)
• Make docs' header levels increase consecutively (#1540,@rzetelskik)
• Generate API docs (#1541,@tnozicka)
• Mark 1.11 as stable version in documentation (#1543,@zimnx)
• Update release schedule (#1544,@zimnx)
• Update support matrix (#1545,@zimnx)
• Add Scylla Operator demo (#1560,@tnozicka)
• Update outdated hyperlink to GKE's raw block devices docs in GKE docs (#1586,@rzetelskik)
• Set kubelet's CPU manager policy for ScyllaDB node pool only in GKE docs and examples (#1595,@rzetelskik)
• Fix invalid commands in Scylla Manager section of GKE mulit-datacenter documentation (#1596,@rzetelskik)
• Add examples for popular platforms regarding LoadBalancer service exposing (#1601,@zimnx)
• Fix typo in helm docs (#1627,@smyja)
• Update documentation about restoring from backup (#1678,@zimnx)
• Add zendesk tag for docs (#1681,@dgarcia360)
• Fix headings in 'Deploying Scylla stack using Helm Charts' doc page (#1750,@rzetelskik)
• Update version requirements for Scylla Manager (#1777,@rzetelskik)
• Remove TLS disablement note from restore docs (#1833,@rzetelskik)
• Document using ScyllaDB APIs (#1834,@tnozicka)
• Document using Alternator API (#1843,@tnozicka)
• Update restore procedure docs with a workaround for Scylla OS 5.4.X and ScyllaDB Enterprise 2024.1.X (#1844,@rzetelskik)

Flaky/Failing Test

• Reenable E2E test for shard-aware ports (#1459,@zimnx)
• Avoid controller-gen issue with path parsing (#1547,@tnozicka)
• Limit scylla manager API calls duration (#1615,@tnozicka)
• Allow Ingress Controller to notice Ingress objects in connection via Ingress E2E (#1631,@zimnx)
• Extend ScyllaCluster rollout timeout in E2E (#1636,@zimnx)
• Wait for e2e Pod even if logs end prematurely (#1699,@tnozicka)
• Fix kubectl logs retry in gke e2e script (#1700,@tnozicka)

Other

• Replace recommonmark with MyST-Parser in docs (#1418,@rzetelskik)
• Enable orphaned PV e2e test on envs with multiple Nodes (#1443,@zimnx)
• Update ScyllaDB Documentation Theme to 1.6 (#1460,@dgarcia360)
• Bump Go version to 1.21 (#1470,@zimnx)
• Use the same tag version of gen-release-notes when preparing them (#1472,@zimnx)
• Add 'externalSeeds' support to ScyllaDB Helm chart (#1487,@rzetelskik)
• Catch up poetry to 1.5 and fix path detection (#1489,@tnozicka)
• Bump k8s-local-volume-provisioner to 0.2.0-rc.0 in CI (#1501,@zimnx)
• Update release procedure (#1506,@zimnx)
• Use correct key in error message when getting agent token (#1521,@tnozicka)
• Fix GH Action installing kubernetes (#1522,@zimnx)
• Generalise naming of networking's parsing helpers (#1524,@rzetelskik)
• Move aggregated deploy manifests to deploy folder (#1532,@tnozicka)
• Make 1.11 docs use MyST (#1535,@rzetelskik)
• Update GitHub issue templates (#1549,@tnozicka)
• Fix a duplicate id in GitHub 'Feature request' issue template (#1550,@rzetelskik)
• Fix GitHub issue template formatting (#1551,@tnozicka)
• Generate aggregated third-party manifests (#1552,@tnozicka)
• Fix crd tmpdir location in Makefile (#1553,@tnozicka)
• Add ScyllaCluster and scylla config examples (#1564,@tnozicka)
• Generelize GKE NodeConfig so it can be faked on libvirt (#1565,@tnozicka)
• Fix scylladb example so it deploys on default gke nodes (#1567,@tnozicka)
• Check for presence of ARTIFACTS env var in e2e script (#1584,@rzetelskik)
• Trap unblocking functions in e2e script's subprocesses (#1585,@rzetelskik)
• Fix Operator deployment in GKE and EKS example scripts (#1597,@rzetelskik)
• Remove image publishing in favour of Prow (#1616,@tnozicka)
• Move e2e namespace initialization before Operator deployment (#1625,@zimnx)
• Fix default ServiceAccount awaiting in E2E framework (#1634,@zimnx)
• Fix workouround for controler-gen interpreting numbers in file paths (#1646,@tnozicka)
• Update local volume provisioner for CI (#1652,@tnozicka)
• Wait for kube-root-ca.crt when creating e2e namespace (#1658,@tnozicka)
• Unify waiting for object state (#1659,@tnozicka)
• Replace project-specific scylladb_markdown extension with sphinx-scylladb-markdown extension (#1661,@dgarcia360)
• Add proposal for specifying ScyllaCluster metadata (#1665,@tnozicka)
• Add Apply functions for corev1.Endpoints and discoveryv1.EndpointSlices (#1703,@zimnx)
• Move ScyllaDB main config file generation from sidecar to operator (#1713,@tnozicka)
• Extend alternator proposal with using Authorization by default (#1731,@tnozicka)
• Pin tags to 1.12 (#1753,@tnozicka)
• Use real salted_hash in alternator e2e (#1835,@tnozicka)