release v1.4.0-beta

.github/workflows/images-ci.yaml

@@ -0,0 +1,36 @@
+name: CI for build images
+
+on:
+  pull_request:
+    branches:
+      - main
+
+# Environment variables available to all jobs and steps in this workflow.
+env:
+  REGISTRY: registry.cn-zhangjiakou.aliyuncs.com
+  NAMESPACE: polardbx-ci
+  TAG: ${{ github.sha }}
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+
+      # 1 Setup go environment
+      - name: Setup go environment
+        uses: actions/setup-go@v3
+        with:
+          go-version: '1.18'
+
+      # 2 Build images
+      - name: Build images
+        run: |-
+          make build REPO=$REGISTRY/$NAMESPACE TAG=$TAG
+
+      # 3 Generate helm chart
+      - name: Generate helm chart
+        run: |-
+          make helm-package
+

.gitignore

@@ -554,4 +554,8 @@ vendor
 !LICENSES/vendor
 !test/testcases/inner
 !pkg/debug
-tools/xstore/venv
+tools/xstore/venv
+
+docs/**
+/hack/make-rules/venv/
+/hack/make-rules/.idea/

CHANGELOG/README.md

@@ -1,5 +1,25 @@
 # Changelog
 
+## 2023-03-17
+
+Release v1.4.0-beta
+
++ Enhancement & New Features
+  + Support consensus log backup
+  + Support point-in-time recovery(PITR)
+  + Support backup schedule
+  + Support restore without backup object
+  + Support rebuild dn follower automatically when `Slave_SQL_Running` becomes `No` due to `Last_SQL_Error`
+  + Support debug runmode of PolarDB-X cluster
+  + Support default PrometheusRule for polardbx cluster
+  + Use *Cluster IP* as the default communication address among dn pods
+  + Use container network as the default network type
++ Bug Fix
+  + Fix backup failure on follower when using non-host network
+  + Fix grafana cpu and mem usage issue for kube-state-metrics 2.3.0
+  + Fix restore issue when binlog checksum is none
+  + Fix cdc crash during backup caused by sending heartbeat using XA transaction
+
 ## 2022-11-17
 
 Release v1.3.0
@@ -61,6 +81,7 @@ Fix overriding my.cnf passing to the script without section.
 Release v1.2.1.
 
 This is a bugfix release.
+
 + Fix the CDC problem by setting the 'consensuslog_revise' variable of galaxyengine to ON.
 + Update docker registry mirror used in China mainland from ustc to sjtug.
 + Fix the scale in/out progress: use "schedule rebalance" command.
@@ -70,11 +91,19 @@ This is a bugfix release.
 Release v1.2.0.
 
 + Enhancement & New Features
-  + Provide a new CR `PolarDBXMonitor` for declaring the monitoring of some `PolarDBXCluster`. The controller will create `ServiceMonitors` to make prometheus scraping the metrics.
-  + Provide a new chart `polardbx-monitor` which packages a customized [kube-prometheus](https://github.com/prometheus-operator/kube-prometheus) release with predefined dashboards for monitoring PolarDB-X clusters.
-  + Provide support for xpaxos version of [galaxyengine](https://github.com/ApsaraDB/galaxyengine). Now the `PolarDBXCluster` controller will create a typical paxos cluster (leader + follower + logger) for each GMS and DN by default. 
-    + **Note** this is a breaking change. After the upgrade, the old `XStores` will be in an unmaintainable state. Update of the GMS/DN is not possible due to incompatible data/log formats. You may have to delete all the `PolarDBXCluster` in your Kubernetes before/after the upgrade.
-    + If you want to keep the compatibility, you can disable this feature by declaring feature gate with negative symbol `EnableGalaxyCluster-`. After that, no multi-node `XStore` with galaxy engine can be created.
+  + Provide a new CR `PolarDBXMonitor` for declaring the monitoring of some `PolarDBXCluster`. The controller will
+    create `ServiceMonitors` to make prometheus scraping the metrics.
+  + Provide a new chart `polardbx-monitor` which packages a
+    customized [kube-prometheus](https://github.com/prometheus-operator/kube-prometheus) release with predefined
+    dashboards for monitoring PolarDB-X clusters.
+  + Provide support for xpaxos version of [galaxyengine](https://github.com/ApsaraDB/galaxyengine). Now
+    the `PolarDBXCluster` controller will create a typical paxos cluster (leader + follower + logger) for each GMS and
+    DN by default.
+    + **Note** this is a breaking change. After the upgrade, the old `XStores` will be in an unmaintainable state.
+      Update of the GMS/DN is not possible due to incompatible data/log formats. You may have to delete all
+      the `PolarDBXCluster` in your Kubernetes before/after the upgrade.
+    + If you want to keep the compatibility, you can disable this feature by declaring feature gate with negative
+      symbol `EnableGalaxyCluster-`. After that, no multi-node `XStore` with galaxy engine can be created.
   + Support scaling up/down and self-healing of the `XStores`.
 
 + Bug Fix
@@ -85,7 +114,7 @@ Release v1.2.0.
 Release v1.1.0.
 
 + Enhancement & New Features
-  + Support scaling in/out the PolarDBXCluster. 
+  + Support scaling in/out the PolarDBXCluster.
   + Support configure SSL on PolarDBXCluster.
   + Add a new CRD PolarDBXCluster for setting and reviewing the configs of PolarDB-X cluster.
     + `config.dynamic.CN` is not going to sync with cluster while phase is running.
@@ -94,21 +123,22 @@ Release v1.1.0.
   + Add label "polardbx/name" to Services、Secrets and ConfigMaps owned by PolarDBXCluster.
   + Support webhooks for PolarDBXCluster and PolarDBXClusterKnobs.
   + Support the complete spec of node selectors in `spec.topology.rules` of PolarDBXCluster.
-  + Create headless services for pods of xstore. Record DNS domains instead of raw IP for records of DN in GMS and among galaxyengine xstores. 
+  + Create headless services for pods of xstore. Record DNS domains instead of raw IP for records of DN in GMS and
+    among galaxyengine xstores.
   + Support overwrite image tag in values.yaml (helm).
   + Support collect metrics for hotspot JVM 11.
   + Add e2e test tests.
 
 + Bug Fix
-  + Fix the wrong call stack when logging with `flow.Error` in some cases. 
+  + Fix the wrong call stack when logging with `flow.Error` in some cases.
   + Fix the wrong timeout in polardbx-init.
   + Fix configuring host path of data volumes in values.yaml (helm).
   + Fix removing ini keys in galaxyengine's config.
   + Fix a `removeNull` in hsperfdata.
 
-## 2021-10-15 
+## 2021-10-15
 
-Release v1.0.0. 
+Release v1.0.0.
 
 + Provide v1 APIs:
   + PolarDBXCluster

README.md

@@ -11,7 +11,7 @@ Follow the [快速开始](https://doc.polardbx.com/quickstart/topics/quickstart-
 
 ## Documentation
 
-Please refer to the [CHANGELOG](./CHANGELOG.md) for the notable changes for each version.**
+Please refer to the [CHANGELOG](./CHANGELOG/README.md) for the notable changes for each version.**
 
 Refer to the documentations for more details, such as CRD definitions and operation guides.
 

api/v1/backup_storage_provider.go → api/v1/polardbx/backup_storage.go

@@ -14,7 +14,12 @@ See the License for the specific language governing permissions and
 limitations under the License.
 */
 
-package v1
+package polardbx
+
+import (
+	"errors"
+	"github.com/alibaba/polardbx-operator/pkg/hpfs/filestream"
+)
 
 // BackupStorageProvider defines the configuration of storage for storing backup files.
 type BackupStorageProvider struct {
@@ -33,3 +38,29 @@ const (
 	OSS  BackupStorage = "oss"
 	SFTP BackupStorage = "sftp"
 )
+
+// BackupStorageFilestreamAction records filestream actions related to specified backup storage
+type BackupStorageFilestreamAction struct {
+	Download filestream.Action
+	Upload   filestream.Action
+	List     filestream.Action
+}
+
+func NewBackupStorageFilestreamAction(storage BackupStorage) (*BackupStorageFilestreamAction, error) {
+	switch storage {
+	case OSS:
+		return &BackupStorageFilestreamAction{
+			Download: filestream.DownloadOss,
+			Upload:   filestream.UploadOss,
+			List:     filestream.ListOss,
+		}, nil
+	case SFTP:
+		return &BackupStorageFilestreamAction{
+			Download: filestream.DownloadSsh,
+			Upload:   filestream.UploadSsh,
+			List:     filestream.ListSsh,
+		}, nil
+	default:
+		return nil, errors.New("invalid storage: " + string(storage))
+	}
+}

api/v1/polardbx/config.go

@@ -47,10 +47,15 @@ type CNConfig struct {
 }
 
 type DNConfig struct {
-	MycnfOverwrite    string          `json:"mycnfOverwrite,omitempty"`
-	LogPurgeInterval  metav1.Duration `json:"logPurgeInterval,omitempty"`
-	EnableAuditLog    bool            `json:"enableAuditLog,omitempty"`
-	LogDataSeparation bool            `json:"logDataSeparation,omitempty"`
+	MycnfOverwrite    string                        `json:"mycnfOverwrite,omitempty"`
+	LogPurgeInterval  metav1.Duration               `json:"logPurgeInterval,omitempty"`
+	EnableAuditLog    bool                          `json:"enableAuditLog,omitempty"`
+	LogDataSeparation bool                          `json:"logDataSeparation,omitempty"`
+	Envs              map[string]intstr.IntOrString `json:"envs,omitempty"`
+}
+
+type CDCConfig struct {
+	Envs map[string]intstr.IntOrString `json:"envs,omitempty"`
 }
 
 type Config struct {
@@ -59,4 +64,7 @@ type Config struct {
 
 	// DN config.
 	DN DNConfig `json:"dn,omitempty"`
+
+	// CDC config
+	CDC CDCConfig `json:"cdc,omitempty"`
 }

api/v1/polardbx/restore.go

@@ -17,13 +17,21 @@ limitations under the License.
 package polardbx
 
 type RestoreSpec struct {
-	// BackupSet defines the source of backup set
+	// BackupSet defines the source of backup set.
+	// It works only when PolarDBXBackup object of this BackupSet still exists.
 	BackupSet string `json:"backupset,omitempty"`
 
-	// From defines the source information, either backup sets, snapshot or an running cluster.
+	// StorageProvider defines storage used to perform backup
+	// +optional
+	StorageProvider *BackupStorageProvider `json:"storageProvider,omitempty"`
+
+	// From defines the source information, either a running cluster, backup set path or backup selector.
+	//
+	// If PolarDBXRestoreFrom.BackupSetPath provided, restore will be performed using metadata backup in remote storage.
+	// It works only when BackupSet is empty and StorageProvider is provided.
 	From PolarDBXRestoreFrom `json:"from,omitempty"`
 
-	// Time defines the specified time of the restored data, in the format of 'yyyy-MM-dd HH:mm:ss'. Required.
+	// Time defines the specified time of the restored data, in the format of 'yyyy-MM-ddTHH:mm:ssZ'. Required.
 	Time string `json:"time,omitempty"`
 
 	// TimeZone defines the specified time zone of the restore time. Default is the location of current cluster.
@@ -37,15 +45,32 @@ type RestoreSpec struct {
 	// use default spec, but replicas of dn will be forced to sync with original cluster now. Default is false
 	// +optional
 	SyncSpecWithOriginalCluster bool `json:"syncSpecWithOriginalCluster,omitempty"`
+
+	// BinlogSource defines the binlog datasource
+	// +optional
+	BinlogSource *RestoreBinlogSource `json:"binlogSource,omitempty"`
 }
 
 // PolarDBXRestoreFrom defines the source information of the restored cluster.
 type PolarDBXRestoreFrom struct {
-	// PolarBDXName defines the the polardbx name that this polardbx is restored from. Optional.
+	// PolarBDXName defines the polardbx name that this polardbx is restored from. Optional.
 	// +optional
 	PolarBDXName string `json:"clusterName,omitempty"`
 
+	// BackupSetPath defines the location of backup set in remote storage
+	BackupSetPath string `json:"backupSetPath,omitempty"`
+
 	// BackupSelector defines the selector for the backups to be selected. Optional.
 	// +optional
 	BackupSelector map[string]string `json:"backupSelector,omitempty"`
 }
+
+// RestoreBinlogSource defines the binlog datasource
+type RestoreBinlogSource struct {
+	//Namespace defines the source binlog namespace
+	Namespace string `json:"namespace,omitempty"`
+	//Checksum defines the binlog file checksum.
+	Checksum string `json:"checksum,omitempty"`
+	//StorageProvider defines the source binlog sink
+	StorageProvider *BackupStorageProvider `json:"storageProvider,omitempty"`
+}

api/v1/polardbx/status.go

@@ -18,7 +18,6 @@ package polardbx
 
 import (
 	"fmt"
-
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 )
@@ -154,6 +153,12 @@ type ClusterReplicasStatus struct {
 	CDC *ReplicasStatus `json:"cdc,omitempty"`
 }
 
+// PitrStatus represents the status ot pitr restore
+type PitrStatus struct {
+	PrepareJobEndpoint string `json:"preapreJobEndpoint,omitempty"`
+	Job                string `json:"job,omitempty"`
+}
+
 type MonitorStatus string
 
 const (

api/v1/polardbx/topology.go

@@ -98,7 +98,7 @@ type XStoreTemplate struct {
 	// +optional
 	ServiceType corev1.ServiceType `json:"serviceType,omitempty"`
 
-	// +kubebuilder:default=true
+	// +kubebuilder:default=false
 
 	// HostNetwork mode.
 	// +optional