API Reference

This page is generated from the Go API types by crd-ref-docs. It lists every field; the topic guides show examples and explain behavior.

The API is still v1alpha1, so fields may change while the operator is under active development.

mysql.cnmsql.co/v1alpha1

Package v1alpha1 contains API Schema definitions for the mysql v1alpha1 API group.

Resource Types

• Backup
• BackupList
• Cluster
• ClusterImageCatalog
• ClusterImageCatalogList
• ClusterList
• Database
• DatabaseList
• DatabaseUser
• DatabaseUserList
• ImageCatalog
• ImageCatalogList
• LogicalRestore
• LogicalRestoreList
• ScheduledBackup
• ScheduledBackupList

AffinityConfiguration

AffinityConfiguration contains the info we need to create the affinity rules for Pods.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
enablePodAntiAffinity boolean	Activates anti-affinity for the pods. The operator will define pods anti-affinity unless this field is explicitly set to false		Optional: {}
topologyKey string	TopologyKey to use for anti-affinity configuration. See Kubernetes documentation for more information on how this works		Optional: {}
nodeSelector object (keys:string, values:string)	NodeSelector is map of key-value pairs used to define the nodes on which the pods can run. More info: https://kubernetes.io/docs/concepts/configuration/assign-pod-node/		Optional: {}
nodeAffinity NodeAffinity	NodeAffinity describes node affinity scheduling rules for the pod. More info: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#node-affinity		Optional: {}
tolerations Toleration array	Tolerations is a list of Tolerations that should be set for all the pods, in order to allow them to run on tainted nodes. More info: https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/		Optional: {}
podAntiAffinityType string	PodAntiAffinityType allows the user to decide whether pod anti-affinity between cluster instances should be considered a strong requirement ("required") during scheduling or not ("preferred", default).	preferred	Enum: [preferred required] Optional: {}
additionalPodAntiAffinity PodAntiAffinity	AdditionalPodAntiAffinity allows to specify pod anti-affinity terms to be added to the ones generated by the operator if EnablePodAntiAffinity is set to true (default) or to be used exclusively if set to false.		Optional: {}
additionalPodAffinity PodAffinity	AdditionalPodAffinity allows to specify pod affinity terms to be passed to all the cluster's pods.		Optional: {}

Backup

Backup is a one-shot request to back up a Cluster with XtraBackup, a volume snapshot, or a logical dump. Deleting it leaves the backup files in place unless it carries the mysql.cnmsql.co/cleanup-backup-files finalizer.

Appears in:

• BackupList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	Backup
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec BackupSpec	spec defines the desired state of Backup		Required: {}
status BackupStatus	status defines the observed state of Backup		Optional: {}

BackupConfiguration

BackupConfiguration describes the continuous archiving target for the cluster.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
objectStore S3ObjectStore	ObjectStore is the S3-compatible destination for backups and binlog archiving.		Optional: {}
retentionPolicy string	RetentionPolicy is a duration string (e.g. "30d", "8w") describing how long to keep backups.		Pattern: ^[1-9][0-9]*[dwm]$ Optional: {}
reclaimPolicy BackupReclaimPolicy	ReclaimPolicy controls what happens to the cluster's entire object-store archive (every base backup, the archived binlogs, and the archive index) when the Cluster is deleted. With "Delete" the operator adds a cleanup finalizer and wipes the archive on teardown; with "Retain" (the default) the archive is kept so a deleted Cluster can still be recovered.	Retain	Enum: [Retain Delete] Optional: {}
target BackupTarget	Target instance to take backups from, defaults to a standby if available.	prefer-standby	Enum: [primary prefer-standby] Optional: {}
xtrabackupOptions string array	XtrabackupOptions are extra flags passed to xtrabackup.		Optional: {}
logicalOptions string array	LogicalOptions are extra flags passed to the dump client (mysqldump / mariadb-dump) for logical backups. A Backup's spec.logical.extraArgs replaces them. The operator does not validate them: flags that change the output format or GTID handling break restore.		Optional: {}
jobTemplate BackupJobTemplate	JobTemplate is the default shaping applied to backup worker Jobs created for this cluster: resources, scheduling (nodeSelector/tolerations/affinity/ priorityClassName), extra labels/annotations, and the finished-Job TTL. A per-Backup spec.jobTemplate overrides it field by field. During recovery the resources from this template are also applied to the restore Job.		Optional: {}
continuousArchiving ContinuousArchivingConfiguration	ContinuousArchiving configures continuous binary-log archiving to the object store, the foundation for point-in-time recovery. Disabled by default.		Optional: {}

BackupJobTemplate

BackupJobTemplate is a curated subset of the pod configuration applied to the backup, restore and instance bootstrap worker Jobs. It deliberately does not expose a full PodTemplateSpec: the operator owns the bootstrap init container, the scratch/TLS volumes and mounts, the worker command and args, and the object-store credential env, and a free-form template would let those be broken. Every field is optional; a per-Backup template overrides the cluster-wide spec.backup.jobTemplate field by field.

Appears in:

• BackupConfiguration
• BackupSpec
• LogicalRestoreSpec
• ScheduledBackupSpec

Field	Description	Default	Validation
ttl Duration	TTL is how long the finished backup worker Job is kept before Kubernetes garbage-collects it (its ttlSecondsAfterFinished). When unset on both the Backup and the cluster, the operator keeps the Job for 24h. A zero duration deletes the Job as soon as it finishes.		Optional: {}
activeDeadline Duration	ActiveDeadline bounds how long the backup worker Job may run (its activeDeadlineSeconds). A worker still running past it is killed and the Backup fails with reason DeadlineExceeded, so a stalled upload or source stream surfaces as a failure instead of a Backup left running forever. When unset on both the Backup and the cluster, the operator uses 24h. A zero duration disables the deadline.		Optional: {}
resources ResourceRequirements	Resources sets the resource requests and limits on the backup worker container. Streaming xbstream can be memory-hungry, so operators often want explicit limits. During recovery the cluster-level template's requests/limits also apply to the restore Job. The other instance bootstrap Jobs (initdb, join, import) always use spec.resources.		Optional: {}
nodeSelector object (keys:string, values:string)	NodeSelector constrains the backup worker Job's pod to nodes with matching labels, e.g. to keep backups off the critical nodes. Not applied to instance bootstrap Jobs: they follow the instance's scheduling so the data volume binds where the instance can run.		Optional: {}
tolerations Toleration array	Tolerations allow the backup worker Job's pod to schedule onto tainted nodes. Added to the instance's tolerations on instance bootstrap Jobs.		Optional: {}
affinity Affinity	Affinity sets the affinity/anti-affinity rules for the backup worker Job's pod. Not applied to instance bootstrap Jobs: they follow the instance's scheduling so the data volume binds where the instance can run.		Optional: {}
priorityClassName string	PriorityClassName sets the pod priority for the backup worker Job. Also applies to instance bootstrap Jobs, over spec.priorityClassName.		Optional: {}
labels object (keys:string, values:string)	Labels are merged onto the generated Job and its pod template. Operator labels take precedence on conflict.		Optional: {}
annotations object (keys:string, values:string)	Annotations are merged onto the generated Job and its pod template.		Optional: {}

BackupList

BackupList contains a list of Backup.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	BackupList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items Backup array

BackupMethod

Underlying type: string

BackupMethod is the method used to take a backup.

Validation:

• Enum: [xtrabackup volumeSnapshot logical]

Appears in:

• BackupSpec
• BackupStatus
• ScheduledBackupSpec

Field	Description
xtrabackup	BackupMethodXtrabackup uses Percona XtraBackup to stream a physical backup to the object store.
volumeSnapshot	BackupMethodVolumeSnapshot uses CSI volume snapshots.
logical	BackupMethodLogical takes a SQL dump of the application schemas with the engine's dump client (mysqldump / mariadb-dump). A logical backup is never a base for recovery or point-in-time replay; load it into a new cluster instead.

BackupPhase

Underlying type: string

BackupPhase is the current phase of a Backup.

Appears in:

• BackupStatus

Field	Description
pending	BackupPhasePending means the backup has not started yet.
running	BackupPhaseRunning means the backup is in progress.
completed	BackupPhaseCompleted means the backup finished successfully.
failed	BackupPhaseFailed means the backup failed.

BackupReclaimPolicy

Underlying type: string

BackupReclaimPolicy describes what happens to an object-store archive when the Kubernetes object that owns it (a Backup or a Cluster) is deleted.

Validation:

• Enum: [Retain Delete]

Appears in:

• BackupConfiguration
• BackupSpec
• ScheduledBackupSpec

Field	Description
Retain	BackupReclaimRetain keeps object-store artifacts after the owning object is deleted. This is the default: removing a Kubernetes object never destroys the only copy of a recovery point unless the user opts in.
Delete	BackupReclaimDelete removes the object-store artifacts when the owning object is deleted, via the cleanup finalizer.

BackupSpec

BackupSpec defines the desired state of Backup.

Appears in:

• Backup

Field	Description	Default	Validation
cluster LocalObjectReference	Cluster references the cluster to back up.		Required: {}
objectStore S3ObjectStore	ObjectStore overrides the destination configured on the referenced Cluster. When omitted, the Cluster's backup object store is used.		Optional: {}
method BackupMethod	Method is the backup method to use.	xtrabackup	Enum: [xtrabackup volumeSnapshot logical] Optional: {}
target BackupTarget	Target instance to take the backup from.	prefer-standby	Enum: [primary prefer-standby] Optional: {}
online boolean	Online, when true, performs a non-blocking (hot) backup. Defaults to true.	true	Optional: {}
reclaimPolicy BackupReclaimPolicy	ReclaimPolicy controls what happens to this backup's object-store archive (the backup payload and its manifest) when the Backup object is deleted. With "Delete" the operator adds the cleanup finalizer and removes the archive on deletion; with "Retain" (the default) the archive is kept.	Retain	Enum: [Retain Delete] Optional: {}
jobTemplate BackupJobTemplate	JobTemplate shapes the backup worker Job for this backup: resources, scheduling (nodeSelector/tolerations/affinity/priorityClassName), extra labels/annotations, and the finished-Job TTL. It overrides the cluster-wide spec.backup.jobTemplate field by field.		Optional: {}
logical LogicalBackupOptions	Logical configures the dump when method is "logical".		Optional: {}

BackupStatus

BackupStatus defines the observed state of Backup.

Appears in:

• Backup

Field	Description	Default	Validation
phase BackupPhase	Phase is the current phase of the backup.		Optional: {}
instanceName string	InstanceName is the instance the backup was taken from.		Optional: {}
method BackupMethod	Method is the method that was used.		Enum: [xtrabackup volumeSnapshot logical] Optional: {}
backupId string	BackupID is a unique identifier of the backup in the object store.		Optional: {}
jobName string	JobName is the Kubernetes Job running this backup.		Optional: {}
destinationPath string	DestinationPath is the full path of the backup in the object store.		Optional: {}
objectStore S3ObjectStore	ObjectStore records the destination the backup was uploaded to, resolved at backup time from the Backup spec or the referenced Cluster. It is snapshotted so the cleanup finalizer can still locate and remove the archive after the referenced Cluster is gone.		Optional: {}
binlogObjectStore S3ObjectStore	BinlogObjectStore records the cluster's binary-log archive store when the backup ran. Point-in-time recovery from this backup replays binlogs from it. Unset on backups taken without continuous archiving or before this field existed, in which case the archive is looked up in ObjectStore.		Optional: {}
sha256 string	SHA256 is the checksum of the uploaded backup artifact.		Optional: {}
beginGTID string	BeginGTID/EndGTID record the GTID range covered by the backup. For a physical backup EndGTID is its anchor, the GTID position of its consistent point, which point-in-time recovery replays from. For a logical backup both hold the dump's snapshot GTID (MariaDB only), for reference.		Optional: {}
endGTID string			Optional: {}
beginBinlog string	BeginBinlog/EndBinlog record the binary log coordinates. For a logical backup both hold the dump's snapshot position (file:position), for reference.		Optional: {}
endBinlog string			Optional: {}
databases string array	Databases lists the schemas in a logical backup, which are the ones an import can select.		Optional: {}
startedAt Time	StartedAt/StoppedAt record the backup timing.		Optional: {}
stoppedAt Time			Optional: {}
error string	Error holds the error message if the backup failed.		Optional: {}
conditions Condition array	Conditions represent the latest observations of the backup state.		Optional: {}

BackupTarget

Underlying type: string

BackupTarget describes which instance a backup is taken from.

Validation:

• Enum: [primary prefer-standby]

Appears in:

• BackupConfiguration
• BackupSpec
• ScheduledBackupSpec

Field	Description
primary	BackupTargetPrimary takes backups from the primary instance.
prefer-standby	BackupTargetPreferStandby prefers a standby instance, falling back to the primary if no standby is available.

BootstrapConfiguration

BootstrapConfiguration describes how the cluster is initialised.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
initdb BootstrapInitDB	InitDB bootstraps a fresh, empty cluster.		Optional: {}
recovery BootstrapRecovery	Recovery bootstraps the cluster by restoring a physical backup.		Optional: {}

BootstrapImport

BootstrapImport selects the logical backup a new cluster loads. Exactly one of Backup and Source is set.

Appears in:

• BootstrapInitDB

Field	Description	Default	Validation
backup LocalObjectReference	Backup references a completed logical Backup in this namespace.		Optional: {}
source string	Source is the name of an entry in ExternalClusters whose objectStore holds the dump. The entry's name is the S3 key prefix to find it under. Mutually exclusive with Backup.		Optional: {}
backupID string	BackupID selects a dump under Source. When empty, the latest completed dump is used. Only valid with Source.		Optional: {}
databases string array	Databases loads only these schemas from the dump. Empty loads every schema in it. Each one must be in the dump.		MaxItems: 256 items:MaxLength: 64 items:MinLength: 1 Optional: {}
postImportSQL string array	PostImportSQL is a list of SQL statements run as root after the dump is loaded.		Optional: {}

BootstrapInitDB

BootstrapInitDB configures a fresh cluster initialisation.

Appears in:

• BootstrapConfiguration

Field	Description	Default	Validation
database string	Database is the name of the application database to create.		Optional: {}
owner string	Owner is the name of the application user that owns the database.		Optional: {}
secret LocalObjectReference	Secret references the credentials for the application user. If empty, a secret is generated.		Optional: {}
postInitSQL string array	PostInitSQL is a list of SQL statements run as root after the database is created.		Optional: {}
characterSet string	Encoding/charset of the application database.		Optional: {}
collation string	Collation of the application database.		Optional: {}
import BootstrapImport	Import loads a logical backup (a SQL dump) into the new cluster after it is initialised. The dump must come from the same flavor, and it can come from any supported server series. Replicas then clone the loaded primary.		Optional: {}

BootstrapRecovery

BootstrapRecovery configures bootstrapping from a physical backup.

Appears in:

• BootstrapConfiguration

Field	Description	Default	Validation
backup LocalObjectReference	Backup references a Backup object to recover from.		Optional: {}
source string	Source is the name of an entry in ExternalClusters to recover from. Mutually exclusive with Backup. The entry's objectStore holds the backups and its name is the S3 key prefix to discover them under.		Optional: {}
backupID string	BackupID narrows recovery to a specific base backup within the object store. Only meaningful when Source is set; when empty, the latest completed backup is selected. Ignored when Backup is set.		Optional: {}
recoveryTarget RecoveryTarget	RecoveryTarget describes the point-in-time recovery target. When omitted, recovery proceeds to the latest available point.		Optional: {}

CatalogImage

CatalogImage maps a MySQL series to a container image. The series, not the integer major, is the upgrade unit: MySQL 8.0 and 8.4 are distinct upgrade targets that both live under integer major 8, so a catalog keyed by integer major could not express the 8.0 -> 8.4 hop.

Appears in:

• ImageCatalogSpec

Field	Description	Default	Validation
series string	Series is the MySQL release series in "major.minor" form (e.g. "8.0", "8.4", "9.7"). It must match the image's server version line.		Pattern: ^[0-9]+\.[0-9]+$ Required: {}
image string	Image is the fully qualified Percona Server for MySQL image reference.		Required: {}

CertificatesConfiguration

CertificatesConfiguration contains the needed configurations to handle server and client certificates for TLS and mTLS communication.

Appears in:

• CertificatesStatus
• ClusterSpec

Field	Description	Default	Validation
serverCASecret string	The secret containing the Server CA certificate. If not defined, a new secret will be created with a self-signed CA and will be used to generate the TLS certificate ServerTLSSecret.		Optional: {}
serverTLSSecret string	The secret of type kubernetes.io/tls containing the server TLS certificate and key that will be set as ssl-cert and ssl-key. Should be signed by the CA in ServerCASecret. Its common name must not be cnmsql-operator.		Optional: {}
replicationTLSSecret string	The secret of type kubernetes.io/tls containing the client certificate to authenticate as the replication user. Should be signed by the CA in ClientCASecret. It is the operator's client certificate for the instance control API, so its common name must be cnmsql-operator.		Optional: {}
clientCASecret string	The secret containing the Client CA certificate. If not defined, a new secret will be created with a self-signed CA and will be used to generate all the client certificates.		Optional: {}
serverAltDNSNames string array	The list of additional Subject Alternative Names (SANs) to be added to the server certificate generated by the operator.		Optional: {}

CertificatesStatus

CertificatesStatus contains configuration certificates and related expiration dates.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
serverCASecret string	The secret containing the Server CA certificate. If not defined, a new secret will be created with a self-signed CA and will be used to generate the TLS certificate ServerTLSSecret.		Optional: {}
serverTLSSecret string	The secret of type kubernetes.io/tls containing the server TLS certificate and key that will be set as ssl-cert and ssl-key. Should be signed by the CA in ServerCASecret. Its common name must not be cnmsql-operator.		Optional: {}
replicationTLSSecret string	The secret of type kubernetes.io/tls containing the client certificate to authenticate as the replication user. Should be signed by the CA in ClientCASecret. It is the operator's client certificate for the instance control API, so its common name must be cnmsql-operator.		Optional: {}
clientCASecret string	The secret containing the Client CA certificate. If not defined, a new secret will be created with a self-signed CA and will be used to generate all the client certificates.		Optional: {}
serverAltDNSNames string array	The list of additional Subject Alternative Names (SANs) to be added to the server certificate generated by the operator.		Optional: {}
expirations object (keys:string, values:string)	Expiration dates for all certificates.		Optional: {}

Cluster

Cluster describes a MySQL or MariaDB deployment: its topology, image, storage, bootstrap method, backups, and operational policy.

Appears in:

• ClusterList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	Cluster
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec ClusterSpec	spec defines the desired state of Cluster		Required: {}
status ClusterStatus	status defines the observed state of Cluster		Optional: {}

ClusterImageCatalog

ClusterImageCatalog maps each server series to an instance image, like ImageCatalog, but is cluster-scoped so Clusters in any namespace can use it.

Appears in:

• ClusterImageCatalogList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ClusterImageCatalog
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec ImageCatalogSpec	spec defines the desired state of ClusterImageCatalog		Required: {}

ClusterImageCatalogList

ClusterImageCatalogList contains a list of ClusterImageCatalog.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ClusterImageCatalogList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items ClusterImageCatalog array

ClusterList

ClusterList contains a list of Cluster.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ClusterList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items Cluster array

ClusterMonitoringTLSConfig

ClusterMonitoringTLSConfig configures TLS for cluster metrics scraping.

Appears in:

• MonitoringConfiguration

Field	Description	Default	Validation
enabled boolean	Enabled serves metrics over TLS when true.	false	Optional: {}

ClusterSpec

ClusterSpec defines the desired state of Cluster.

Appears in:

• Cluster

Field	Description	Default	Validation
description string	Description of this MySQL cluster.		Optional: {}
flavor Flavor	Flavor selects the database engine flavor (mysql or mariadb). Immutable after creation. Defaults to mysql.	mysql	Enum: [mysql mariadb] Optional: {}
inheritedMetadata EmbeddedObjectMetadata	Metadata that will be inherited by all objects related to the Cluster.		Optional: {}
imageName string	ImageName is the name of the Percona Server for MySQL container image to use. Mutually exclusive with ImageCatalogRef.		Optional: {}
imageCatalogRef ImageCatalogRef	ImageCatalogRef resolves the image from an ImageCatalog or ClusterImageCatalog based on the MySQL series. Mutually exclusive with ImageName.		Optional: {}
imagePullPolicy PullPolicy	ImagePullPolicy is the policy used to pull the container image.		Enum: [Always Never IfNotPresent] Optional: {}
imagePullSecrets LocalObjectReference array	ImagePullSecrets is the list of pull secrets used to pull the image.		Optional: {}
instances integer	Instances is the number of MySQL instances (one primary + replicas).	1	Minimum: 1 Optional: {}
minSyncReplicas integer	MinSyncReplicas is the minimum number of semi-synchronous replicas that must acknowledge a transaction before it is committed on the primary. MariaDB always waits for exactly one acknowledgement, so on that flavor values above 1 have no effect.		Minimum: 0 Optional: {}
maxSyncReplicas integer	MaxSyncReplicas is the maximum number of semi-synchronous replicas the primary will wait for. Must be lower than the number of instances. On MariaDB, values above 1 have no effect.		Minimum: 0 Optional: {}
mysql MySQLConfiguration	MySQL holds the engine configuration (my.cnf parameters, replication options).		Optional: {}
replication ReplicationConfiguration	Replication selects and tunes the replication topology (asynchronous / semi-synchronous GTID replication, or quorum-based Group Replication). The mode is immutable after creation; when omitted the cluster is async.		Optional: {}
storage StorageConfiguration	Storage configuration for the instance data directory.		Required: {}
binlogStorage StorageConfiguration	BinlogStorage, when set, places the binary logs on a separate volume from the data directory.		Optional: {}
bootstrap BootstrapConfiguration	Bootstrap describes how the cluster is initialised (fresh init or recovery from a backup).		Optional: {}
rootPasswordSecret LocalObjectReference	RootPasswordSecret references a secret containing the password for the MySQL root user. If not provided, a secret is generated.		Optional: {}
enableSuperuserAccess boolean	EnableSuperuserAccess, when true, makes the root user reachable through the generated/provided secret. Defaults to false.	false	Optional: {}
certificates CertificatesConfiguration	Certificates configures the TLS/mTLS material used by the cluster.		Optional: {}
resources ResourceRequirements	Resources describes the compute resource requirements of the instance pods.		Optional: {}
affinity AffinityConfiguration	Affinity/anti-affinity rules for the instance pods.		Optional: {}
topologySpreadConstraints TopologySpreadConstraint array	TopologySpreadConstraints describes how the instance pods should be spread across topology domains.		Optional: {}
priorityClassName string	PriorityClassName for the instance pods.		Optional: {}
schedulerName string	SchedulerName to use for the instance pods.		Optional: {}
primaryUpdateStrategy PrimaryUpdateStrategy	PrimaryUpdateStrategy controls whether the operator performs the primary update automatically (unsupervised) or waits for the user (supervised).	unsupervised	Enum: [unsupervised supervised] Optional: {}
primaryUpdateMethod PrimaryUpdateMethod	PrimaryUpdateMethod controls how the primary is updated: by switchover (promoting a replica first) or by in-place restart.	switchover	Enum: [switchover restart] Optional: {}
inPlaceInstanceManagerUpdates boolean	InPlaceInstanceManagerUpdates, when true, rolls an operator upgrade out to this cluster's instances by streaming the new instance-manager binary to each Pod, which re-execs in place — no Pod restart and no switchover. When false (the default) the operator instead deletes and recreates each Pod one at a time (replicas first, primary last via switchover).		Optional: {}
upgrade UpgradeConfiguration	Upgrade tunes MySQL server major-version upgrades.		Optional: {}
maxStartDelay integer	MaxStartDelay is the time in seconds allowed for an instance to start.	3600	Optional: {}
maxStopDelay integer	MaxStopDelay is the time in seconds allowed for an instance to gracefully shut down.	1800	Optional: {}
smartShutdownTimeout integer	SmartShutdownTimeout is the time in seconds reserved for a "smart" (graceful) shutdown attempt before falling back to a "fast" shutdown. Must be lower than maxStopDelay; the remaining time is used for the fast/immediate fallback. Defaults to 180.		Minimum: 0 Optional: {}
maxSwitchoverDelay integer	MaxSwitchoverDelay is the time in seconds allowed for a switchover to complete before being considered failed.	3600	Optional: {}
failoverDelay integer	FailoverDelay is the amount of time in seconds the operator waits before declaring an unreachable primary failed and triggering a failover.	0	Optional: {}
enableFailover boolean	EnableFailover, when true (default), lets the operator promote a replica when the primary becomes unreachable. Set it to false to pin the primary role to whichever instance currently holds it: the operator promotes nothing, the failed primary's Pod is still recreated so it can recover in place, and the cluster reports Blocked until the primary is back or a manual switchover moves the role. Planned handoffs are unaffected — manual switchover and enableSwitchoverOnDrain keep working. Group Replication clusters ignore this switch: the group elects its own primary.	true	Optional: {}
enablePrimaryLease boolean	EnablePrimaryLease, when true (default), makes the acting primary hold a per-cluster Lease before accepting writes.	true	Optional: {}
failoverPolicy FailoverPolicy	FailoverPolicy bounds and steers automatic primary election. When it is unset, the operator promotes the most advanced reachable replica and puts no limit on how far behind that replica may be.		Optional: {}
backup BackupConfiguration	Backup configures continuous archiving and the object store target.		Optional: {}
replica ReplicaClusterConfiguration	Replica turns this cluster into a replica cluster that follows a source defined in ExternalClusters.		Optional: {}
externalClusters ExternalCluster array	ExternalClusters is the list of external clusters that can be used as a replication source or a recovery origin.		Optional: {}
managed ManagedConfiguration	Managed describes the resources (roles, services) managed declaratively by the operator.		Optional: {}
monitoring MonitoringConfiguration	Monitoring configuration.		Optional: {}
nodeMaintenanceWindow NodeMaintenanceWindow	NodeMaintenanceWindow defines if the cluster is tolerant to node failures during maintenance (e.g. PVC may be reused).		Optional: {}
enablePDB boolean	EnablePDB, when true (default), makes the operator create a PodDisruptionBudget for the cluster.	true	Optional: {}
enableSwitchoverOnDrain boolean	EnableSwitchoverOnDrain, when true (default), makes the operator promote a healthy replica via a planned switchover when the primary Pod is gracefully terminated (e.g. a node drain or eviction), instead of waiting for the primary to become unreachable and failing over. Falls back to failover when no safe candidate exists or the handoff cannot complete in time.	true	Optional: {}
serviceAccountTemplate ServiceAccountTemplate	ServiceAccountTemplate to use for the generated service account.		Optional: {}
env EnvVar array	Env is a list of additional environment variables added to the instance containers.		Optional: {}
envFrom EnvFromSource array	EnvFrom is a list of sources to populate environment variables in the instance containers.		Optional: {}
podSecurityContext PodSecurityContext	PodSecurityContext applied to the instance pods.		Optional: {}
securityContext SecurityContext	SecurityContext applied to the instance containers.		Optional: {}
logLevel string	LogLevel sets the operator-side log level for this cluster.	info	Enum: [error warning info debug trace] Optional: {}

ClusterStatus

ClusterStatus defines the observed state of Cluster.

Appears in:

• Cluster

Field	Description	Default	Validation
flavor Flavor	Flavor echoes the resolved engine flavor (spec.flavor, default mysql).		Enum: [mysql mariadb] Optional: {}
instances integer	Instances is the total number of instances reported.		Optional: {}
readyInstances integer	ReadyInstances is the number of ready instances.		Optional: {}
labelSelector string	LabelSelector is the serialized label selector that identifies the instance Pods managed by this Cluster. It is populated by the operator and exposed through the scale sub-resource (selectorpath) so that autoscalers such as the Vertical Pod Autoscaler can discover the managed Pods to resize.		Optional: {}
instanceNames string array	InstanceNames is the list of instance (pod) names.		Optional: {}
currentPrimary string	CurrentPrimary is the name of the instance currently acting as primary.		Optional: {}
targetPrimary string	TargetPrimary is the name of the instance that should become primary (used during switchover/failover).		Optional: {}
currentPrimaryTimestamp Time	CurrentPrimaryTimestamp is when the current primary was elected.		Optional: {}
currentPrimaryGeneration integer	CurrentPrimaryGeneration counts the changes of currentPrimary: the instance that records itself as currentPrimary raises it by one in the same update (the operator does under Group Replication). It fences the binlog archive: a primary stamps its generation into the archive index, and a writer with a lower generation, a demoted primary still finishing a pass, never judges the archive's segments.		Optional: {}
targetPrimaryTimestamp Time	TargetPrimaryTimestamp is when the current switchover request to TargetPrimary was started. It bounds the switchover by spec.maxSwitchoverDelay.		Optional: {}
divergedInstances string array	DivergedInstances are replicas whose executed GTID set is not contained in the primary's (errant transactions). They cannot safely rejoin; their in-Pod reconciler reads this list and refuses to self-configure as a replica, leaving them read-only for an operator to resolve.		Optional: {}
fencedInstances string array	FencedInstances are instances the operator has fenced because their Pod carries the fencing annotation. A fenced instance is pulled out of the routing Services, kept read-only by its in-Pod reconciler (so it stops archiving and refuses writes), and is excluded as a failover candidate. Clearing the annotation removes it from this list and restores it.		Optional: {}
failedInstances string array	FailedInstances are instances whose Pod shows positive evidence of being unable to run: a Failed Pod phase, or a container stuck in CrashLoopBackOff after repeated restarts. Unlike a not-yet-ready instance (which is expected during initial provisioning), a failed instance is a degradation regardless of whether the cluster ever finished provisioning, so it is surfaced independently of the phase. It is the cluster's "unhealthy" bucket.		Optional: {}
replicationBrokenInstances string array	ReplicationBrokenInstances are reachable replicas whose replication has aborted with a recorded error — a stopped IO or SQL thread, e.g. a duplicate-key conflict that halts replication. Unlike a diverged instance (detected by comparing GTID sets and listed in DivergedInstances), this is derived from the SQL-layer replication error the in-Pod reconciler reports, so a replica that is Running but cannot replicate is surfaced as a degradation rather than being mistaken for one still finishing provisioning.		Optional: {}
resizingPVC string array	ResizingPVC lists the PVCs whose storage is currently being expanded — the volume-level resize or the node-side filesystem grow has not yet completed. For a backend that cannot expand a volume in use (storage.resizeInUseVolumes false) a name lingers here until the operator recycles the owning Pod and the fresh mount finishes the resize.		Optional: {}
primaryFailingSince Time	PrimaryFailingSince records when the current primary first became unreachable. It is used to enforce spec.failoverDelay before an automatic failover, and is cleared once the primary is healthy again.		Optional: {}
primaryHealthySince Time	PrimaryHealthySince records when the current primary began the unbroken stretch of health it is now in: it is stamped the first time the primary is observed healthy and re-stamped whenever it recovers from a failing stretch, so it moves forward every time the primary drops out. It is what spec.failoverPolicy.primaryStabilityWindow is measured from. Never earlier than CurrentPrimaryTimestamp, since a primary's health cannot predate its election.		Optional: {}
lastFailoverTimestamp Time	LastFailoverTimestamp records when the operator last promoted a replica in an automatic failover. It is what spec.failoverPolicy.minTimeBetweenFailovers is measured from. A switchover, planned or requested, does not set it: the cooldown exists to damp the operator reacting to failures, not to delay a promotion somebody asked for.		Optional: {}
latestGeneratedNode integer	LatestGeneratedNode is the serial of the latest generated instance.		Optional: {}
establishedAt Time	EstablishedAt records the first time the cluster reached full readiness (every instance ready together), marking that it completed initial provisioning at least once. It is sticky: once set it is never cleared, so a later degradation cannot reset it. Its presence is what distinguishes a cluster that is still being provisioned (a drop below readiness is expected) from an established one (a drop below readiness is a degradation). It is deliberately independent of Phase, which intermediate reconcile steps re-stamp and which therefore cannot carry this fact reliably.		Optional: {}
phase string	Phase is a high-level human-readable cluster phase.		Optional: {}
phaseReason string	PhaseReason gives more detail about the current phase.		Optional: {}
image string	Image is the resolved image currently in use.		Optional: {}
targetImage ImageInfo	TargetImage is what the operator learned about the image it rolls the instances to, by running it in a probe Pod: the server version and flavor the image's server binary reports, and the digest the kubelet pulled. A new image is probed and validated before any instance moves to it; until then the cluster stays on this one (see the ImageReady condition).		Optional: {}
gtidExecutedByInstance object (keys:string, values:string)	GTIDExecutedByInstance maps an instance name to its gtid_executed set.		Optional: {}
mariadbTimeline MariaDBEpoch array	MariaDBTimeline (MariaDB) records each change of primary, oldest first. Entry i says that, in every replication domain, the transactions after its handoff up to the next entry's handoff were authored by its server id. A MariaDB GTID position names only the author of its last transaction; this history is what tells a forked former primary from a lagging one. The oldest entries are dropped once no instance position or archive segment needs them. Not recorded on replica clusters.		MaxItems: 256 Optional: {}
gtidExecutedUpdatedAt Time	GTIDExecutedUpdatedAt records when GTIDExecutedByInstance was last refreshed. Because gtid_executed advances on every write, the operator throttles how often it persists the map; this timestamp marks the last persisted snapshot.		Optional: {}
replicationLagByInstance object (keys:string, values:integer)	ReplicationLagByInstance maps an instance name to its replication lag in milliseconds, as measured by the heartbeat (see spec.replication.heartbeat). Instances that reported no reading are absent rather than zero, since zero would say "in sync", which is the opposite of what an unreadable heartbeat means. It is here to be looked at. The failover bound does not read it: it asks the surviving replicas for their readings at the moment it elects, which is both fresher than this snapshot and, unlike the departing primary's GTID position, still answerable. Expect these values to climb while a primary is down, because nothing is stamping the heartbeat any more.		Optional: {}
replicationLagUpdatedAt Time	ReplicationLagUpdatedAt records when ReplicationLagByInstance was last refreshed. Lag moves constantly, so the operator throttles how often it persists the map; this timestamp says how old the readings are, and is the only way to tell a replica that is genuinely one second behind from one whose last reading merely happened to say so.		Optional: {}
executableHashByInstance object (keys:string, values:string)	ExecutableHashByInstance maps an instance name to the SHA-256 hash of its running instance manager binary, as reported by the in-Pod control API. The operator uses it to detect stale instance managers when upgrading.		Optional: {}
operatorExecutableHash string	OperatorExecutableHash is the SHA-256 hash of the running operator binary. It is the target hash every instance manager should match after an upgrade.		Optional: {}
certificates CertificatesStatus	Certificates reports the status of the managed certificates.		Optional: {}
continuousArchiving ContinuousArchivingStatus	ContinuousArchiving reports the health of continuous binlog archiving when it is enabled.		Optional: {}
lastRetentionRunTime Time	LastRetentionRunTime is when the operator last ran a backup-retention GC pass against the object store. It throttles the periodic pass.		Optional: {}
dumpAccountSecretVersion string	DumpAccountSecretVersion is the resourceVersion of the <cluster>-dump Secret last applied to the cnmsql_dump account on the primary. The operator re-applies the account when the Secret changes. Written only by the operator.		Optional: {}
dumpAccountServerVersion string	DumpAccountServerVersion is the server version of the primary the cnmsql_dump account was last applied on. Its grants depend on the version, so the operator re-applies the account when the primary's version changes, after an in-place upgrade. Written only by the operator.		Optional: {}
observedGeneration integer	ObservedGeneration is the generation observed by the controller.		Optional: {}
conditions Condition array	Conditions represent the latest available observations of the cluster's state.		Optional: {}
managedRolesStatus ManagedRolesStatus	ManagedRolesStatus reports the reconciliation state of the declarative managed roles.		Optional: {}
groupReplication GroupReplicationStatus	GroupReplication reflects the live group membership and quorum, mirrored from performance_schema.replication_group_members. Nil for async clusters. The operator is the sole writer of this block.		Optional: {}

ConfigMapKeySelector

ConfigMapKeySelector selects a key from a ConfigMap in the same namespace.

Appears in:

• MonitoringConfiguration

Field	Description	Default	Validation
name string	Name of the ConfigMap		Required: {}
key string	Key within the ConfigMap to select		Required: {}

ContinuousArchivingConfiguration

ContinuousArchivingConfiguration configures continuous binary-log (binlog) archiving: the current primary's instance manager ships rotated binlog files to the object store so the cluster keeps a gapless, GTID-addressable change history.

Appears in:

• BackupConfiguration

Field	Description	Default	Validation
enabled boolean	Enabled turns continuous binlog archiving on. Requires Backup.ObjectStore.	false	Optional: {}
targetRPOSeconds integer	TargetRPOSeconds bounds the recovery point objective: the primary forces a binary-log rotation at least this often so a low-write cluster still archives promptly. Defaults to 300 (5 minutes).	300	Minimum: 10 Optional: {}
maxBinlogSizeMB integer	MaxBinlogSizeMB caps the active binary log before mysqld rotates it, bounding the size-based RPO and per-object size. Defaults to 16 MiB.	16	Minimum: 1 Optional: {}
binlogExpireSeconds integer	BinlogExpireSeconds is how long mysqld keeps a binary log on the instance's data volume before expiring it. With the active purge gate off (the default) this is the only mechanism that reclaims binlog space, so it also sets the window in which a lagged or returning replica can still catch up from the primary's logs. Budget disk for it: the data volume must hold this many seconds of binlog writes on top of the dataset. 0 selects the default. Defaults to 604800 (7 days).	604800	Minimum: 0 Optional: {}
purgeAfterArchive boolean	PurgeAfterArchive turns on the active purge gate: the primary runs PURGE BINARY LOGS on a binary log once it is safely in the object store and every other instance of the cluster has applied the transactions it holds, reclaiming binlog space without waiting for binlogExpireSeconds. The replicas' positions come from status.gtidExecutedByInstance, which the operator refreshes at least every five minutes, so purging trails writes by about that much. An instance whose position is unknown (still joining, for example) holds every file; a fenced instance holds the files it has not applied; a diverged instance, which must be re-cloned anyway, holds none. When the same file stays held for a while the BinlogPurgeHeld condition names the instances holding it. binlogExpireSeconds still applies on its own, so a replica down for longer than that must be re-cloned. Defaults to false.	false	Optional: {}
objectStore S3ObjectStore	ObjectStore is where the binary-log archive is written. When unset, the archive goes to spec.backup.objectStore next to the base backups. The archive keeps the same layout in either store: <path>/<cluster>/binlogs/. Changing it starts a new archive in the new store from the oldest binary log still on the primary; take a new base backup afterwards.		Optional: {}

ContinuousArchivingStatus

ContinuousArchivingStatus reports the health and frontier of continuous binary-log archiving.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
enabled boolean	Enabled mirrors whether continuous archiving is configured on.		Optional: {}
destination string	Destination is the archive location in use, <endpoint>/<bucket>/<path> (endpoint empty for AWS). A change is reported with an ArchiveMoved Warning event.		Optional: {}
lastArchivedBinlog string	LastArchivedBinlog is the most recent binary-log file shipped by the current primary.		Optional: {}
lastArchivedGTID string	LastArchivedGTID is the last GTID covered by the archive.		Optional: {}
lastArchivedTime Time	LastArchivedTime is when the most recent file finished archiving.		Optional: {}
pendingFiles integer	PendingFiles is the number of rotated binary logs not yet archived (archive lag). A growing value means the archiver is falling behind.		Optional: {}
lastFailureReason string	LastFailureReason and LastFailureTime record the most recent archiving failure on the current primary, if any.		Optional: {}
lastFailureTime Time			Optional: {}
purgeHeldBy string array	PurgeHeldBy lists the instances that have not applied the oldest archived binary log on the primary, which keeps the purge gate from removing it. Empty when the purge gate is off or holds nothing back.		Optional: {}
purgeHeldSince Time	PurgeHeldSince is when the purge gate started keeping that binary log.		Optional: {}
forkGTIDs string array	ForkGTIDs lists, one entry per archive segment, the archived transactions the surviving timeline never executed (a dead branch left by a lagged promotion): the MySQL GTID set, or a MariaDB range such as 0-1-219..0-1-225. Point-in-time recovery to a time or to the latest point leaves them out; an explicit targetGTID can still recover them.		Optional: {}
forkDetectedAt Time	ForkDetectedAt is when the earliest of those records was written.		Optional: {}
oldestSegmentPosition string	OldestSegmentPosition (MariaDB) is the lowest GTID position any archive segment reached, per domain. The operator keeps the MariaDB primary timeline back to it.		Optional: {}
disownedGTIDs string	DisownedGTIDs (MySQL) is every transaction the archive recorded as disowned, including dead branches whose segment retention already dropped and the dead part of base backups. An instance holding any of them is diverged.		Optional: {}
gaps string array	Gaps lists the stretches of the timeline the archive is missing between transactions it holds (a MySQL GTID set, or a MariaDB domain-first..last range): point-in-time recovery from a base backup taken before one of them cannot cross it. They come from a binary log that left the server before it was archived, typically a successor cloned after the last archived file whose predecessor then died.		Optional: {}
gapsSince Time	GapsSince is when the archive started reporting these gaps. A gap the former primary's drain fills soon after a failover never raises ArchiveGap; one that outlives the grace period does.		Optional: {}
coveredGTIDSet string	CoveredGTIDSet (MySQL) is every transaction the archive holds. Failover prefers, among equally advanced replicas, one whose binary-log history the archive already covers.		Optional: {}

Database

Database declares a MySQL schema and the accounts scoped to it on a Cluster in the same namespace. The controller issues only the SQL needed to converge and drops nothing unless asked to.

Appears in:

• DatabaseList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	Database
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec DatabaseSpec	spec defines the desired state of Database		Required: {}
status DatabaseStatus	status defines the observed state of Database		Optional: {}

DatabaseGrant

DatabaseGrant describes a single MySQL GRANT statement.

Appears in:

• InlineUser

Field	Description	Default	Validation
privileges string array	Privileges is the list of privileges (e.g. "SELECT", "INSERT", "ALL").		Required: {}
on string	On is the target of the grant (e.g. "mydb.*"). Defaults to the managed database.		Optional: {}

DatabaseList

DatabaseList contains a list of Database.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	DatabaseList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items Database array

DatabaseSpec

DatabaseSpec defines the desired state of Database.

Appears in:

• Database

Field	Description	Default	Validation
cluster LocalObjectReference	Cluster references the MySQL cluster this database belongs to.		Required: {}
name string	Name is the name of the MySQL database (schema). Defaults to the resource name if empty.		Optional: {}
ensure EnsureOption	Ensure controls whether the database is created or dropped.	present	Enum: [present absent] Optional: {}
characterSet string	CharacterSet of the database (e.g. "utf8mb4").		Optional: {}
collation string	Collation of the database (e.g. "utf8mb4_0900_ai_ci").		Optional: {}
users InlineUser array	Users is the list of users managed for this database.		Optional: {}
reclaimPolicy string	ReclaimPolicy controls what happens to the MySQL database when the Database object is deleted.	retain	Enum: [delete retain] Optional: {}
driftDetection boolean	DriftDetection controls whether the operator periodically re-applies this database and its declared users to correct out-of-band drift. When true (the default) the controller re-asserts the desired state on every resync, so a manually dropped schema or altered user is restored within readyResync. When false the controller reconciles only on changes to the spec or to a referenced password Secret, and does not periodically re-apply; drift introduced out of band is not corrected until the next such change.	true	Optional: {}

DatabaseStatus

DatabaseStatus defines the observed state of Database.

Appears in:

• Database

Field	Description	Default	Validation
applied boolean	Applied is true once the desired state has been reconciled.		Optional: {}
message string	Message provides additional detail, typically an error.		Optional: {}
observedGeneration integer	ObservedGeneration is the generation observed by the controller.		Optional: {}
passwordStatus object (keys:string, values:string)	PasswordStatus records, per managed user ("name@host"), the source Secret resourceVersion last applied to MySQL. It lets the controller re-apply a password only when its Secret changes.		Optional: {}
conditions Condition array	Conditions represent the latest observations of the database state.		Optional: {}

DatabaseUser

DatabaseUser declares one installation-wide MySQL account (name@host) and its grants on a Cluster in the same namespace. Unlike the inline users of a Database, it is not scoped to a single schema.

Appears in:

• DatabaseUserList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	DatabaseUser
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec DatabaseUserSpec	spec defines the desired state of DatabaseUser		Required: {}
status DatabaseUserStatus	status defines the observed state of DatabaseUser		Optional: {}

DatabaseUserGrant

DatabaseUserGrant describes a single MySQL GRANT statement.

Appears in:

• DatabaseUserSpec

Field	Description	Default	Validation
privileges string array	Privileges is the list of privileges (e.g. "SELECT", "INSERT", "ALL").		MinItems: 1
on string	On is the target of the grant (e.g. ".", "mydb.", "mydb.mytable"). Defaults to ".*".	.	Optional: {}

DatabaseUserList

DatabaseUserList contains a list of DatabaseUser.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	DatabaseUserList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items DatabaseUser array

DatabaseUserRevoke

DatabaseUserRevoke describes a single MySQL REVOKE statement, applied after the grants. The target is required: a revoke must name the schema (or schema object) it carves out, so it cannot accidentally strip a global grant.

Appears in:

• DatabaseUserSpec

Field	Description	Default	Validation
privileges string array	Privileges is the list of privileges to revoke (e.g. "INSERT", "UPDATE").		MinItems: 1
on string	On is the target to revoke from (e.g. "mysql.", "sys.").		Required: {}

DatabaseUserSpec

DatabaseUserSpec defines the desired state of DatabaseUser. A DatabaseUser owns a single installation-wide MySQL account (name@host); unlike the inline users of a Database, it is not scoped to a single schema and its grants may target anything in the instance.

Appears in:

• DatabaseUser

Field	Description	Default	Validation
cluster LocalObjectReference	Cluster references the MySQL cluster this user belongs to.		Required: {}
name string	Name is the MySQL user name. Defaults to the resource name if empty.		MaxLength: 32 Optional: {}
host string	Host the user connects from. Defaults to "%".	%	Optional: {}
ensure EnsureOption	Ensure controls whether the user is created or dropped.	present	Enum: [present absent] Optional: {}
passwordSecret SecretKeySelector	PasswordSecret references the secret key holding the user's password. Required when Ensure=present.		Optional: {}
superuser boolean	Superuser grants ALL PRIVILEGES on . WITH GRANT OPTION. It is unsafe for multi-tenant use and is mutually exclusive with Grants. For a constrained "DBaaS admin", use a Grants entry with ALL on . instead, which omits WITH GRANT OPTION.	false	Optional: {}
maxUserConnections integer	MaxUserConnections resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxQueriesPerHour integer	MaxQueriesPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxUpdatesPerHour integer	MaxUpdatesPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxConnectionsPerHour integer	MaxConnectionsPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
requireTLS string	RequireTLS sets REQUIRE X509, REQUIRE SSL, or none.	none	Enum: [x509 ssl none] Optional: {}
comment string	Comment is an optional user comment.		Optional: {}
grants DatabaseUserGrant array	Grants is the list of grants applied to the user. Targets are installation-wide and have no default schema (unlike Database.spec.users).		Optional: {}
revokes DatabaseUserRevoke array	Revokes lists privileges to revoke after Grants are applied, each scoped to an explicit target (typically a system schema such as "mysql."). Combined with partial_revokes=ON on the cluster, this carves the system schemas out of a broad ".*" grant so a cross-database admin cannot modify the grant tables and self-escalate. Revokes are applied after Grants, in order. Not supported on MariaDB: it has neither partial_revokes nor DENY (DENY is planned for MariaDB 13.1), so the carve-out cannot be enforced. A DatabaseUser with revokes on a MariaDB cluster is rejected at admission and stays unapplied (reason UnsupportedOnMariaDB).		Optional: {}
reclaimPolicy string	ReclaimPolicy controls what happens to the MySQL user when the DatabaseUser object is deleted.	retain	Enum: [delete retain] Optional: {}
driftDetection boolean	DriftDetection controls whether the operator periodically re-applies this user to correct out-of-band drift. When true (the default) the controller re-asserts the desired state on every resync, so a manually altered or dropped account is restored within readyResync. When false the controller reconciles only on changes to the spec or to the referenced password Secret, and does not periodically re-apply; drift introduced out of band is not corrected until the next such change.	true	Optional: {}

DatabaseUserStatus

DatabaseUserStatus defines the observed state of DatabaseUser.

Appears in:

• DatabaseUser

Field	Description	Default	Validation
applied boolean	Applied is true once the desired state has been reconciled.		Optional: {}
message string	Message provides additional detail, typically an error.		Optional: {}
observedGeneration integer	ObservedGeneration is the generation observed by the controller.		Optional: {}
passwordSecretResourceVersion string	PasswordSecretResourceVersion records the source Secret resourceVersion last applied to MySQL, so the password is re-applied only when it changes.		Optional: {}
conditions Condition array	Conditions represent the latest observations of the user state.		Optional: {}

EmbeddedObjectMetadata

EmbeddedObjectMetadata contains metadata to be inherited by all resources related to a Cluster.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
labels object (keys:string, values:string)			Optional: {}
annotations object (keys:string, values:string)			Optional: {}

EnsureOption

Underlying type: string

EnsureOption controls whether a declarative object must be present or absent.

Validation:

• Enum: [present absent]

Appears in:

• DatabaseSpec
• DatabaseUserSpec
• InlineUser
• RoleConfiguration

Field	Description
present	EnsurePresent means the object must exist.
absent	EnsureAbsent means the object must not exist.

ExternalCluster

ExternalCluster describes a MySQL server external to this Cluster, used as a replication source or recovery origin.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
name string	Name of the external cluster, referenced by Replica.Source and BootstrapRecovery.Source.		Required: {}
connectionParameters object (keys:string, values:string)	ConnectionParameters is a key/value map of connection settings (host, port, etc.).		Optional: {}
password SecretKeySelector	Password references a secret key holding the password for the connection.		Optional: {}
sslCert SecretKeySelector	SSLCert/SSLKey/SSLRootCert reference secret keys for mTLS to the source.		Optional: {}
sslKey SecretKeySelector			Optional: {}
sslRootCert SecretKeySelector			Optional: {}
objectStore S3ObjectStore	ObjectStore allows recovering from a backup stored in an object store.		Optional: {}
binlogObjectStore S3ObjectStore	BinlogObjectStore is where the external cluster's binary-log archive lives, when it is not in ObjectStore. Recovery reads base backups from ObjectStore and binlogs from here.		Optional: {}

FailoverPolicy

FailoverPolicy bounds and steers the operator's automatic election of a new primary.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
maxTransactionsBehind integer	MaxTransactionsBehind caps how many transactions a replica may be missing and still be promoted during an unplanned failover. The operator measures the gap in GTIDs, against the last position it recorded for the primary in status.gtidExecutedByInstance, because the primary itself cannot be asked once it is down. It refreshes that snapshot periodically rather than on every commit, so the gap is a lower bound on the true loss and can under report it, but never over report it. The gap is not read from Seconds_Behind_Source. That column reports how far the SQL applier trails the events it has already received. A replica that stopped receiving reads zero there once it drains its relay log, no matter how many transactions the primary committed without it, which is precisely the case that loses data. The column is also NULL whenever the replication IO thread is disconnected, and it always is on a replica whose primary has just failed. If every replica exceeds the bound, the operator refuses the failover and moves the cluster to Blocked, reporting the size of the gap, rather than promote a replica and discard the transactions it never received. Writes then stay down until the old primary comes back with its data, or until an operator accepts the loss by raising this bound. Leave it unset for no bound, in which case the most advanced replica is promoted however far behind it is.		Minimum: 0 Optional: {}
maxReplicationLag Duration	MaxReplicationLag caps how far behind a replica may be, in seconds of writes, and still be promoted during an unplanned failover. It states the same bound as MaxTransactionsBehind in the unit a recovery objective is normally written in: a hundred transactions may be a millisecond of writes or an hour of them, and only the clock version can be held against an RPO. The measurement comes from the heartbeat the primary stamps into a replicated table (see spec.replication.heartbeat), which must be enabled for this bound to have anything to read. A replica's raw reading climbs by one second per second once the primary stops stamping, so the operator subtracts how long the primary has been failing and is left with the delay as it stood when the primary died. That subtraction starts from the moment the operator noticed the failure, which is never earlier than the failure itself, so the result overstates the loss rather than understating it. It is checked only against replicas that actually missed transactions. A replica holding every transaction the primary committed loses nothing by being promoted, however far its applier has fallen behind, and the wait for it to drain its relay log is delay rather than loss. If no replica is within the bound, the operator refuses the failover and moves the cluster to Blocked, exactly as it does for MaxTransactionsBehind. Leave it unset for no time bound.		Optional: {}
preferredPrimary string array	PreferredPrimary names the instances that should hold the primary role, most preferred first, as a way to steer the primary onto particular nodes: pin an instance to a node with the storage or the zone you want writes to land on, and name it here. It steers two things. In an election it orders the candidates, so the most preferred one that is also safe wins. It is only ever a tie-break: a preferred instance that has diverged, that is missing transactions the bounds above refuse to lose, or that cannot be proven to hold every other candidate's transactions, is passed over for one that can. Preference never promotes a replica that the operator would otherwise have refused. It also brings the primary back. When the primary is not the most preferred instance that is currently available, and that instance is a healthy replica fit to be a switchover target, the operator performs a planned switchover to it. That happens after any failover away from it has healed, so a preferred instance that keeps failing does not cause the primary to bounce back to it: the failback waits out MinTimeBetweenFailovers like any other automatic promotion. Names are instance (Pod) names, e.g. "my-cluster-1". Instances that do not exist are ignored, so a list may name instances of a cluster larger than the current one. Leave it unset to let the election pick purely on safety and ordinal order, and to never move a healthy primary.		Optional: {}
minTimeBetweenFailovers Duration	MinTimeBetweenFailovers damps a flapping primary: it refuses an automatic failover that comes less than this long after the previous one settled, so a primary that dies, is replaced, and whose replacement dies again does not walk the role through every instance in the cluster in a matter of minutes. A refusal moves the cluster to Blocked and names the time remaining. It does not stop the operator from recreating the failed primary's Pod, which is the outcome the wait exists to allow: a primary that is failing intermittently is usually better restarted in place than replaced. It bounds automatic promotions only. A switchover you request by writing status.targetPrimary is honoured immediately, since a human asking for the primary to move is not the operator flapping. Leave it unset to allow a failover as soon as spec.failoverDelay has elapsed, however recently the last one happened.		Optional: {}
primaryStabilityWindow Duration	PrimaryStabilityWindow decides when a promoted primary counts as settled, and so when MinTimeBetweenFailovers starts running. Without it the cooldown runs from the moment of the promotion, and a primary that comes up, flaps, and comes up again still burns the cooldown down while it is misbehaving; once the cooldown expires the operator fails away from an instance that was never really working. With it, the cooldown runs from the point the primary had been continuously healthy for this long. Every time it drops out, the clock restarts, so an instance that cannot stay up never settles and the operator keeps declining to fail over on its behalf, leaving it to recover or an operator to intervene rather than dragging the rest of the cluster through the churn. A primary that has been healthy for longer than this window settled long ago, which is every primary in a cluster that is not flapping, so this costs nothing in the ordinary case. A promoted primary that never became healthy at all cannot settle by staying up, so the cooldown falls back to running from the promotion and the operator is never trapped waiting on an instance that will not arrive. It has no effect on its own without MinTimeBetweenFailovers, which is the bound it qualifies.		Optional: {}

Flavor

Underlying type: string

Flavor selects the database engine a Cluster runs.

Validation:

• Enum: [mysql mariadb]

Appears in:

• ClusterSpec
• ClusterStatus
• ImageCatalogSpec
• ImageInfo

Field	Description
mysql
mariadb

GroupMember

GroupMember is one member's state within the group.

Appears in:

• GroupReplicationStatus

Field	Description	Default	Validation
instance string	Instance is the pod name of the member.
state string	State is the member's group state: ONLINE, RECOVERING, OFFLINE, ERROR or UNREACHABLE.
role string	Role is the member's group role: PRIMARY or SECONDARY.
reachable boolean	Reachable reports whether the group currently considers this member reachable (not UNREACHABLE).

GroupReplicationConfiguration

GroupReplicationConfiguration tunes MySQL Group Replication. All fields map to group_replication_* server variables; the operator owns the rest of the namespace (group name, addresses, seeds, start/bootstrap control).

Appears in:

• ReplicationConfiguration

Field	Description	Default	Validation
consistency string	Consistency maps to group_replication_consistency, the transaction consistency guarantee the group enforces.	BEFORE_ON_PRIMARY_FAILOVER	Enum: [EVENTUAL BEFORE_ON_PRIMARY_FAILOVER BEFORE AFTER BEFORE_AND_AFTER] Optional: {}
exitStateAction string	ExitStateAction maps to group_replication_exit_state_action: what a member does when it involuntarily leaves the group (e.g. on an unrecoverable error or loss of quorum).	READ_ONLY	Enum: [READ_ONLY OFFLINE_MODE ABORT_SERVER] Optional: {}
autoRejoinTries integer	AutoRejoinTries maps to group_replication_autorejoin_tries: how many times a member tries to automatically rejoin after being expelled before giving up.	3	Minimum: 0 Optional: {}
groupName string	GroupName, when set, pins group_replication_group_name (a UUID). When unset the operator generates one on first bootstrap and persists it to status.groupReplication.groupName. Immutable once the group exists; a changed group name fractures the group.		Optional: {}

GroupReplicationStatus

GroupReplicationStatus is the operator's cross-validated view of the group, aggregated from every member's reported replication_group_members.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
groupName string	GroupName is the pinned group_replication_group_name (a UUID). Sticky and immutable once set: a changed group name fractures the group.		Optional: {}
bootstrapped boolean	Bootstrapped records that the group has been created at least once. Sticky: false→true on first bootstrap, never auto-cleared. It makes group bootstrap exactly-once across restarts and re-elections; re-arming it is the path to a split-brain second group.		Optional: {}
primaryMember string	PrimaryMember is the pod name of the member the group elected PRIMARY. status.currentPrimary is mirrored from this field.		Optional: {}
members GroupMember array	Members is the per-member view of the group.		Optional: {}
hasQuorum boolean	HasQuorum reports whether a majority of configured members is ONLINE and reachable, i.e. the group can make progress.
observedViewMax integer	ObservedViewMax is the largest group view size the operator has ever observed for this group. Sticky: never decreases. Used as the quorum denominator so that a group that loses members is recognised as quorum-lost, while a bootstrapping group uses its current view size.		Optional: {}
observedOnlineMax integer	ObservedOnlineMax is the largest number of ONLINE members the operator has ever observed for this group. Sticky, tracked alongside ViewMax.		Optional: {}
viewId string	ViewID is the current group view identifier. It changes on every membership change and is one of the signals the operator reconciles on.		Optional: {}
communicationProtocol string	CommunicationProtocol is the effective minimum-compatible protocol reported by group_replication_get_communication_protocol(). It can differ from the server version passed to the setter (MySQL 8.4 reports 8.0.27).		Optional: {}
communicationProtocolTarget string	CommunicationProtocolTarget is the MySQL server version most recently passed successfully to group_replication_set_communication_protocol(). It is the idempotency marker for post-upgrade protocol finalization.		Optional: {}

ImageCatalog

ImageCatalog maps each server series to an instance image for Clusters in its namespace. A Cluster selects an entry with spec.imageCatalogRef.

Appears in:

• ImageCatalogList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ImageCatalog
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec ImageCatalogSpec	spec defines the desired state of ImageCatalog		Required: {}

ImageCatalogList

ImageCatalogList contains a list of ImageCatalog.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ImageCatalogList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items ImageCatalog array

ImageCatalogRef

ImageCatalogRef references an ImageCatalog or ClusterImageCatalog entry to resolve a container image for a given MySQL series.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
apiGroup string	APIGroup is the group for the resource being referenced. If APIGroup is not specified, the specified Kind must be in the core API group. For any other third-party types, APIGroup is required.		Optional: {}
kind string	Kind is the type of resource being referenced
name string	Name is the name of resource being referenced
series string	Series is the MySQL release series to resolve in the catalog, in "major.minor" form (e.g. "8.0", "8.4", "9.7").		Pattern: ^[0-9]+\.[0-9]+$ Required: {}

ImageCatalogSpec

ImageCatalogSpec is the shared spec for ImageCatalog and ClusterImageCatalog.

Appears in:

• ClusterImageCatalog
• ImageCatalog

Field	Description	Default	Validation
flavor Flavor	Flavor is the advisory engine flavor for the series in this catalog (mysql or mariadb). It is non-gating: a mismatch does not block resolution, but improves the admission error message.		Enum: [mysql mariadb] Optional: {}
images CatalogImage array	Images is the list of MySQL series to container image mappings. Each series must appear at most once.		MaxItems: 8 MinItems: 1

ImageInfo

ImageInfo describes an instance image as its own server binary reports it.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
image string	Image is the image reference that was probed, as resolved from spec.imageName or the image catalog.
imageID string	ImageID is the image the kubelet pulled for the probe, as reported in the probe Pod's container status (typically repository@sha256:digest).		Optional: {}
flavor Flavor	Flavor is the engine the image contains (mysql or mariadb).		Enum: [mysql mariadb]
serverVersion string	ServerVersion is the server release the image contains, e.g. "8.4.11".

InlineUser

InlineUser describes a MySQL user managed declaratively as part of a Database. For an installation-wide user not scoped to a single database, use the standalone DatabaseUser CRD instead.

Appears in:

• DatabaseSpec

Field	Description	Default	Validation
name string	Name of the user.		Required: {}
host string	Host the user connects from. Defaults to "%".	%	Optional: {}
ensure EnsureOption	Ensure controls whether the user is created or dropped.	present	Enum: [present absent] Optional: {}
passwordSecret SecretKeySelector	PasswordSecret references the secret key holding the user's password.		Optional: {}
grants DatabaseGrant array	Grants is the list of grants applied to the user.		Optional: {}

LocalObjectReference

LocalObjectReference contains the reference to a Kubernetes object in the same namespace, identified by name.

Appears in:

• BackupSpec
• BootstrapImport
• BootstrapInitDB
• BootstrapRecovery
• ClusterSpec
• DatabaseSpec
• DatabaseUserSpec
• LogicalRestoreSpec
• ScheduledBackupSpec

Field	Description	Default	Validation
name string	Name of the referent		Required: {}

LogicalBackupOptions

LogicalBackupOptions configures a logical (SQL dump) backup.

Appears in:

• BackupSpec
• ScheduledBackupSpec

Field	Description	Default	Validation
databases string array	Databases limits the dump to these schemas. Empty means every application schema. The system schemas (mysql, sys, performance_schema, information_schema) and operator-owned schemas are always excluded.		MaxItems: 256 items:MaxLength: 64 items:MinLength: 1 Optional: {}
extraArgs string array	ExtraArgs are appended to the dump command. They replace the cluster's spec.backup.logicalOptions. The operator does not validate them: flags that change the output format or GTID handling break restore.		Optional: {}

LogicalRestore

LogicalRestore loads selected databases from a logical backup into a running cluster's primary.

Appears in:

• LogicalRestoreList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	LogicalRestore
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec LogicalRestoreSpec	spec defines the desired state of LogicalRestore		Required: {}
status LogicalRestoreStatus	status defines the observed state of LogicalRestore		Optional: {}

LogicalRestoreList

LogicalRestoreList contains a list of LogicalRestore.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	LogicalRestoreList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items LogicalRestore array

LogicalRestorePhase

Underlying type: string

LogicalRestorePhase is the current phase of a LogicalRestore.

Appears in:

• LogicalRestoreStatus

Field	Description
pending	LogicalRestorePhasePending means the restore has not started: the dump or the cluster's primary is not ready yet.
running	LogicalRestorePhaseRunning means the restore worker Job is loading the dump.
completed	LogicalRestorePhaseCompleted means every selected database was loaded.
failed	LogicalRestorePhaseFailed means the restore failed. The status error says whether the selected databases were changed.

LogicalRestorePolicy

Underlying type: string

LogicalRestorePolicy says what a restore does with a selected database that already holds objects.

Validation:

• Enum: [FailIfExists DropAndRecreate]

Appears in:

• LogicalRestoreSpec

Field	Description
FailIfExists	LogicalRestoreFailIfExists refuses the whole restore when a selected database holds a table, view, routine or event. Nothing is changed. An empty database, such as one a Database resource created, is loaded into.
DropAndRecreate	LogicalRestoreDropAndRecreate drops each selected database, then loads it from the dump. Schema-level grants survive the drop.

LogicalRestoreSpec

LogicalRestoreSpec defines the desired state of LogicalRestore. It is immutable: a restore is a one-shot action.

Appears in:

• LogicalRestore

Field	Description	Default	Validation
cluster LocalObjectReference	Cluster is the running cluster to load into. The dump is loaded into its current primary and reaches the replicas through replication.		Required: {}
backup LocalObjectReference	Backup references a completed logical Backup in this namespace. Mutually exclusive with Source.		Optional: {}
source string	Source names an entry of the target Cluster's spec.externalClusters whose object store holds the dump. Mutually exclusive with Backup.		Optional: {}
backupID string	BackupID selects a dump under Source. Empty picks the latest.		Optional: {}
databases string array	Databases are the schemas loaded from the dump. It is required: a restore never loads a whole dump implicitly. Each must be in the dump.		MaxItems: 256 MinItems: 1 items:MaxLength: 64 items:MinLength: 1
policy LogicalRestorePolicy	Policy says what to do with a selected database that already holds objects: FailIfExists refuses the restore, DropAndRecreate drops the database first. It is required, so an overwrite is always explicit.		Enum: [FailIfExists DropAndRecreate] Required: {}
jobTemplate BackupJobTemplate	JobTemplate shapes the restore worker Job: resources, scheduling, extra labels and annotations, the finished-Job TTL and the deadline. It overrides the cluster-wide spec.backup.jobTemplate field by field, like a Backup's.		Optional: {}

LogicalRestoreStatus

LogicalRestoreStatus defines the observed state of LogicalRestore.

Appears in:

• LogicalRestore

Field	Description	Default	Validation
phase LogicalRestorePhase	Phase is the current phase of the restore.		Optional: {}
targetInstance string	TargetInstance is the primary the dump is loaded into.		Optional: {}
jobName string	JobName is the Kubernetes Job running the restore.		Optional: {}
backupID string	BackupID identifies the dump in the object store.		Optional: {}
sourcePath string	SourcePath is the full object-store path of the dump.		Optional: {}
databases string array	Databases lists the schemas restored.		Optional: {}
startedAt Time	StartedAt/StoppedAt record the restore timing.		Optional: {}
stoppedAt Time			Optional: {}
error string	Error holds the error message if the restore failed, and says whether the selected databases were changed.		Optional: {}
conditions Condition array	Conditions represent the latest observations of the restore state.		Optional: {}

ManagedConfiguration

ManagedConfiguration describes resources managed declaratively by the operator.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
services ManagedServices	Services describes the services managed for the cluster.		Optional: {}
roles RoleConfiguration array	Roles is the list of MySQL users (roles) managed declaratively on the primary instance.		Optional: {}

ManagedRoleStatus

Underlying type: string

ManagedRoleStatus is the reconciliation state of a single managed role.

Appears in:

• ManagedRolesStatus

Field	Description
reconciled	ManagedRoleReconciled means the role matches its desired state.
not-managed	ManagedRoleNotManaged means the MySQL user exists but is not managed.
pending-reconciliation	ManagedRolePendingReconciliation means the role still needs work.
reserved	ManagedRoleReserved means the role name is reserved by the operator.

ManagedRolesStatus

ManagedRolesStatus reports the reconciliation state of managed roles.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
byStatus object (keys:ManagedRoleStatus, values:string array)	ByStatus groups managed role names by their reconciliation status.		Optional: {}
cannotReconcile object (keys:string, values:string array)	CannotReconcile maps a role name to the reasons it could not be reconciled.		Optional: {}
passwordStatus object (keys:string, values:RolePasswordState)	PasswordStatus tracks the applied password Secret version per role.		Optional: {}

ManagedService

ManagedService describes a user-defined managed service.

Appears in:

• ManagedServices

Field	Description	Default	Validation
selectorType ServiceSelectorType	SelectorType specifies the type of selectors the service will have. Valid values are "rw", "r", and "ro".		Enum: [rw r ro] Required: {}
name string	Name is the name of the additional service. Must be unique among all managed services and must not collide with the default service names (<cluster>-rw, <cluster>-ro, <cluster>-r).		Required: {}
updateStrategy ServiceUpdateStrategy	UpdateStrategy describes how the service template is reconciled with the operator defaults.	patch	Enum: [patch replace] Optional: {}
serviceTemplate ServiceTemplateSpec	ServiceTemplate is the template specification for the service. When UpdateStrategy is "patch", fields here are merged on top of the role-specific defaults. When "replace", they replace the defaults entirely (except for selector and owner reference).		Optional: {}

ManagedServices

ManagedServices controls the services generated for the cluster.

Appears in:

• ManagedConfiguration

Field	Description	Default	Validation
disabledDefaultServices ServiceSelectorType array	DisabledDefaultServices is the list of default services (rw, ro, r) to disable. The rw service cannot be disabled.		Enum: [rw ro r] Optional: {}
template ServiceTemplateSpec	Template applies to the three default services (rw, ro, r). Fields set here are merged into each default service. The operator still chooses the selector and port based on the service role.		Optional: {}
additional ManagedService array	Additional is a list of additional managed services specified by the user. Each entry declares a selectorType and an optional template to overlay on top of the role-specific defaults.		Optional: {}

MariaDBEpoch

MariaDBEpoch is one change of primary on a MariaDB cluster.

Appears in:

• ClusterStatus

Field	Description	Default	Validation
instance string	Instance is the primary of this epoch.
serverID integer	ServerID is its @@server_id, the server component of the GTIDs it authors.
handoff string	Handoff is its @@gtid_slave_pos when it took authority: per domain, the last transaction it inherited.		Optional: {}
since Time	Since is when the operator first observed it as primary.

Metadata

Metadata is a structure similar to the metav1.ObjectMeta, but still parseable by controller-gen to create a suitable CRD for the user. The comment of PodTemplateSpec has an explanation of why we are not using the core data types.

Appears in:

• ServiceAccountTemplate

Field	Description	Default	Validation
name string	The name of the resource. Only supported for certain types		Optional: {}
labels object (keys:string, values:string)	Map of string keys and values that can be used to organize and categorize (scope and select) objects. May match selectors of replication controllers and services. More info: http://kubernetes.io/docs/user-guide/labels		Optional: {}
annotations object (keys:string, values:string)	Annotations is an unstructured key value map stored with a resource that may be set by external tools to store and retrieve arbitrary metadata. They are not queryable and should be preserved when modifying objects. More info: http://kubernetes.io/docs/user-guide/annotations		Optional: {}

MonitoringConfiguration

MonitoringConfiguration configures cluster monitoring.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
enablePodMonitor boolean	EnablePodMonitor creates a PodMonitor for the cluster pods.	false	Optional: {}
customQueriesConfigMap ConfigMapKeySelector array	CustomQueriesConfigMap references config maps holding custom monitoring queries.		Optional: {}
customQueriesSecret SecretKeySelector array	CustomQueriesSecret references secrets holding custom monitoring queries.		Optional: {}
disableDefaultQueries boolean	DisableDefaultQueries disables the built-in monitoring query set.		Optional: {}
metricsQueriesTTL Duration	MetricsQueriesTTL is the minimum interval between executions of the default and custom monitoring queries.		Optional: {}
privileges RolePrivilege array	Privileges are extra grants for the cnmsql_metrics account that custom queries run as. Only SELECT and SHOW VIEW on a database (db.*) or a table (db.table) are allowed; . and the mysql schema are refused. The operator applies them on the primary and revokes any other grant the account holds, except its built-in ones. Each entry must set on.		MaxItems: 32 Optional: {}
tls ClusterMonitoringTLSConfig	TLS configures TLS for the instance metrics endpoint.		Optional: {}

MySQLConfiguration

MySQLConfiguration holds the MySQL engine configuration.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
parameters object (keys:string, values:string)	Parameters is a key/value map of my.cnf settings applied under the [mysqld] section.		Optional: {}
binlogFormat string	BinlogFormat is the binary log format. ROW is required for safe replication and PITR and is the default.	ROW	Enum: [ROW STATEMENT MIXED] Optional: {}
semiSync SemiSyncConfiguration	SemiSync configures semi-synchronous replication.		Optional: {}
additionalConfigFiles object (keys:string, values:string)	AdditionalConfigFiles are extra files dropped into the MySQL configuration directory, keyed by file name.		Optional: {}

NodeMaintenanceWindow

NodeMaintenanceWindow contains information that the operator will use while upgrading the underlying nodes.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
reusePVC boolean	ReusePVC, when true, reuses the existing PVC during a node maintenance (instead of provisioning a fresh one).	true	Optional: {}
inProgress boolean	InProgress signals that a node maintenance is in progress.	false	Optional: {}

ObjectMetaTemplate

ObjectMetaTemplate carries the user-configurable metadata fields.

Appears in:

• ServiceTemplateSpec

Field	Description	Default	Validation
labels object (keys:string, values:string)	Labels added to the Service.		Optional: {}
annotations object (keys:string, values:string)	Annotations added to the Service.		Optional: {}

PrimaryUpdateMethod

Underlying type: string

PrimaryUpdateMethod contains the method to use when upgrading the primary server of the cluster as part of rolling updates.

Validation:

• Enum: [switchover restart]

Appears in:

• ClusterSpec

Field	Description
switchover	PrimaryUpdateMethodSwitchover means the operator promotes a replica before updating the former primary.
restart	PrimaryUpdateMethodRestart means the operator restarts the primary in place.

PrimaryUpdateStrategy

Underlying type: string

PrimaryUpdateStrategy contains the strategy to follow when upgrading the primary server of the cluster as part of rolling updates.

Validation:

• Enum: [unsupervised supervised]

Appears in:

• ClusterSpec

Field	Description
unsupervised	PrimaryUpdateStrategyUnsupervised means that the operator performs the switchover/restart of the primary automatically.
supervised	PrimaryUpdateStrategySupervised means that the operator waits for the user to manually trigger the primary update.

RecoveryTarget

RecoveryTarget allows to specify a point in time to recover to.

Appears in:

• BootstrapRecovery

Field	Description	Default	Validation
targetTime string	TargetTime is an RFC3339 timestamp to recover to.		Optional: {}
targetGTID string	TargetGTID is the GTID set to recover up to.		Optional: {}
targetImmediate boolean	TargetImmediate stops recovery as soon as a consistent state is reached.		Optional: {}

ReplicaClusterConfiguration

ReplicaClusterConfiguration turns the cluster into a replica that follows an external source.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
enabled boolean	Enabled controls whether the cluster runs in replica mode.	true	Optional: {}
source string	Source is the name of the entry in ExternalClusters to replicate from.		Required: {}

ReplicationConfiguration

ReplicationConfiguration selects and tunes the replication topology.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
mode string	Mode is the replication topology. Immutable after creation.	async	Enum: [async groupReplication] Optional: {}
groupReplication GroupReplicationConfiguration	GroupReplication tunes the group when Mode=groupReplication. It must not be set unless Mode is groupReplication.		Optional: {}
heartbeat ReplicationHeartbeat	Heartbeat tunes the replication-lag heartbeat, which is on by default.		Optional: {}
maxReadyLag Duration	MaxReadyLag is the maximum replication lag a replica may report and still be Ready. It shapes read traffic, not failover: a replica that is further behind than this bound keeps failing its readiness probe, so the Kubernetes endpoints controller holds it out of the -ro and -r Services and the Cluster reports Degraded until it has caught up. This closes the window in which a replica returning from a long outage — or scaled up from a volume cloned hours earlier — serves reads that are hours old while its replication threads are nominally running. The measurement is the heartbeat lag the instance already computes (see spec.replication.heartbeat): the age of the newest stamp it has applied, which grows in wall-clock seconds no matter how fast the applier chases the primary. The heartbeat must be enabled for the bound to have anything to read, so setting this bound while explicitly disabling the heartbeat is rejected. A replica that cannot produce a heartbeat reading at all — before its first stamp arrives, or while the heartbeat read is failing — is treated as over the bound rather than assumed caught up, so it stays out of the read Services until it can prove otherwise. The bound must be at least three heartbeat intervals: the reading grows by up to one interval between stamps even on a caught-up replica. The reading compares the primary's clock (the stamp) with the replica's, so clock skew between their nodes adds to it; keep the bound well above the skew NTP allows. While the primary is down or stops stamping (a failover, a primary that cannot write), every replica's reading grows, so once it passes the bound all replicas leave the read Services until a primary stamps again. It gates async replicas only. Under Group Replication a member is ready only once it is ONLINE, which already excludes a member that is still recovering; leave this unset there. It is also not a failover bound: what a replica may lose when promoted is governed by failoverPolicy.maxReplicationLag and failoverPolicy.maxTransactionsBehind. Leave it unset (or zero) to disable the gate, which is the default: a replica whose replication threads are running is Ready however far behind it is.		Optional: {}

ReplicationHeartbeat

ReplicationHeartbeat configures the replication-lag heartbeat: the writable primary's instance manager stamps the current time into a small replicated table, and every instance reads it back to measure how old the newest write it has caught up to is.

It exists because the server cannot answer that question itself. Seconds_Behind_Source times the SQL applier against the events it has already received, so a replica that stopped receiving anything reports zero once its relay log drains, no matter how far the primary ran on without it, and it reports NULL whenever the IO thread is disconnected, which is the state every replica is in the moment its primary dies.

The table is heartbeat.heartbeat, in pt-heartbeat's shape, so the heartbeat collector already built into mysqld_exporter scrapes it as-is.

Appears in:

• ReplicationConfiguration

Field	Description	Default	Validation
enabled boolean	Enabled turns the heartbeat on. It defaults to true: the cost is one small replicated write per interval on the primary, and it is what failoverPolicy.maxReplicationLag and the lag metrics read.	true	Optional: {}
interval Duration	Interval is how often the primary stamps the table. It sets the floor on every lag reading, because a replica in perfect sync still reports an age of up to one interval simply because the next stamp is not written yet. Keep it well under any maxReplicationLag you set. Defaults to 1s.	1s	Optional: {}

RoleConfiguration

RoleConfiguration describes a MySQL user managed declaratively against the primary instance.

Appears in:

• ManagedConfiguration

Field	Description	Default	Validation
name string	Name is the MySQL user name.		MaxLength: 32 Required: {}
host string	Host is the MySQL host part. Defaults to "%".	%	Optional: {}
ensure EnsureOption	Ensure controls whether the user should exist or be absent.	present	Enum: [present absent] Optional: {}
passwordSecret SecretKeySelector	PasswordSecret references a Secret key holding the user's password. When unset, the operator generates a password and stores it in a Secret named <cluster>-<name> with key "password".		Optional: {}
superuser boolean	Superuser grants ALL PRIVILEGES on . WITH GRANT OPTION.	false	Optional: {}
maxUserConnections integer	MaxUserConnections resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxQueriesPerHour integer	MaxQueriesPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxUpdatesPerHour integer	MaxUpdatesPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
maxConnectionsPerHour integer	MaxConnectionsPerHour resource limit. 0 = no limit.		Minimum: 0 Optional: {}
requireTLS string	RequireTLS sets the connection TLS requirement: x509, ssl, or none.	none	Enum: [x509 ssl none] Optional: {}
privileges RolePrivilege array	Privileges are grants (global or per-database). Mutually exclusive with Superuser.		Optional: {}

RolePasswordState

RolePasswordState records the password Secret version last applied for a role.

Appears in:

• ManagedRolesStatus

Field	Description	Default	Validation
secretResourceVersion string	SecretResourceVersion is the resourceVersion of the password Secret last applied.		Optional: {}
lastApplied Time	LastApplied is when the password was last applied.		Optional: {}

RolePrivilege

RolePrivilege is a grant of one or more privileges on a target.

Appears in:

• MonitoringConfiguration
• RoleConfiguration

Field	Description	Default	Validation
privileges string array	Privileges is the grant list (SELECT, INSERT, ALL, etc.).		MinItems: 1
on string	On is the target (e.g. ".", "mydb."). Defaults to ".*".		Optional: {}

S3Credentials

S3Credentials holds the references to the secrets containing the credentials used to access an S3-compatible object store.

Appears in:

• S3ObjectStore

Field	Description	Default	Validation
accessKeyId SecretKeySelector	The reference to the secret key containing the access key id.		Optional: {}
secretAccessKey SecretKeySelector	The reference to the secret key containing the secret access key.		Optional: {}
sessionToken SecretKeySelector	The reference to the secret key containing the session token, used for temporary credentials.		Optional: {}
inheritFromIAMRole boolean	InheritFromIAMRole, when true, makes the credentials be retrieved from the pod's environment (IRSA / instance profile) instead of from a secret.	false	Optional: {}

S3ObjectStore

S3ObjectStore describes an S3-compatible object store, designed to be compatible with as many providers as possible (AWS S3, MinIO, Ceph RGW, Wasabi, Backblaze B2, etc.).

Appears in:

• BackupConfiguration
• BackupSpec
• BackupStatus
• ContinuousArchivingConfiguration
• ExternalCluster

Field	Description	Default	Validation
endpoint string	Endpoint is the URL of the S3-compatible service. Leave empty to target AWS S3 with the region's default endpoint.		Optional: {}
region string	Region is the bucket region. Required by most providers; for AWS it selects the regional endpoint. When empty it defaults to the region the endpoint expects requests to be signed with: "auto" for Cloudflare R2, "us-east-1" everywhere else (which MinIO, Ceph and friends accept).		Optional: {}
bucket string	Bucket is the destination bucket name.		Required: {}
path string	Path is the key prefix (folder) inside the bucket under which backups are stored.		Optional: {}
forcePathStyle boolean	ForcePathStyle uses path-style addressing (endpoint/bucket/key) instead of virtual-hosted style (bucket.endpoint/key). Required by MinIO, Ceph and most non-AWS providers; defaults to true for maximum compatibility.	true	Optional: {}
signatureVersion S3SignatureVersion	SignatureVersion selects the request signing scheme. Defaults to s3v4; set to s3v2 for legacy providers.	s3v4	Enum: [s3v4 s3v2] Optional: {}
serverSideEncryption string	ServerSideEncryption sets the SSE algorithm applied to every uploaded object: AES256 (SSE-S3), aws:kms, or aws:kms:<key-id>. Leave unset on providers that encrypt at rest unconditionally (MinIO, R2, B2), which reject or silently ignore the SSE header.		Optional: {}
storageClass string	StorageClass sets the storage class of every uploaded object (e.g. "STANDARD_IA"). Leave unset on providers with a single class; a class the provider does not know is rejected at upload time.		Optional: {}
credentials S3Credentials	Credentials to access the object store.		Required: {}
tls S3TLSConfig	TLS configuration for the endpoint connection.		Optional: {}

S3SignatureVersion

Underlying type: string

S3SignatureVersion is the AWS Signature version used to sign object-store requests.

Validation:

• Enum: [s3v4 s3v2]

Appears in:

• S3ObjectStore

Field	Description
s3v4	SignatureVersionV4 is the default AWS Signature V4.
s3v2	SignatureVersionV2 is the legacy AWS Signature V2, kept for older S3-compatible providers.

S3TLSConfig

S3TLSConfig configures TLS verification against the object-store endpoint.

Appears in:

• S3ObjectStore

Field	Description	Default	Validation
insecureSkipVerify boolean	InsecureSkipVerify disables TLS certificate verification against the endpoint. Use only for testing.	false	Optional: {}
caBundleSecret SecretKeySelector	CABundleSecret references a secret key holding a PEM CA bundle used to verify the endpoint certificate (for private CAs / self-signed endpoints).		Optional: {}

ScheduledBackup

ScheduledBackup creates Backup objects for a Cluster on a cron schedule.

Appears in:

• ScheduledBackupList

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ScheduledBackup
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ObjectMeta	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec ScheduledBackupSpec	spec defines the desired state of ScheduledBackup		Required: {}
status ScheduledBackupStatus	status defines the observed state of ScheduledBackup		Optional: {}

ScheduledBackupList

ScheduledBackupList contains a list of ScheduledBackup.

Field	Description	Default	Validation
apiVersion string	mysql.cnmsql.co/v1alpha1
kind string	ScheduledBackupList
kind string	Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds		Optional: {}
apiVersion string	APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources		Optional: {}
metadata ListMeta	Refer to Kubernetes API documentation for fields of metadata.
items ScheduledBackup array

ScheduledBackupSpec

ScheduledBackupSpec defines the desired state of ScheduledBackup.

Appears in:

• ScheduledBackup

Field	Description	Default	Validation
schedule string	Schedule is a cron expression (6 fields, including seconds) defining when backups are taken.		Required: {}
cluster LocalObjectReference	Cluster references the cluster to back up.		Required: {}
suspend boolean	Suspend, when true, pauses the schedule.	false	Optional: {}
immediate boolean	Immediate, when true, takes a backup as soon as the ScheduledBackup is created, in addition to the schedule.	false	Optional: {}
backupOwnerReference string	BackupOwnerReference controls the owner reference set on the generated Backup objects.	self	Enum: [none self cluster] Optional: {}
reclaimPolicy BackupReclaimPolicy	ReclaimPolicy is propagated to every generated Backup as its spec.reclaimPolicy. With "Delete" each generated Backup carries the cleanup finalizer, so deleting it also removes its archive (the backup payload and its manifest) from the object store. Defaults to "Retain", the non-destructive default.	Retain	Enum: [Retain Delete] Optional: {}
method BackupMethod	Method is the backup method used for the generated backups.	xtrabackup	Enum: [xtrabackup volumeSnapshot logical] Optional: {}
target BackupTarget	Target instance to take the generated backups from.	prefer-standby	Enum: [primary prefer-standby] Optional: {}
online boolean	Online, when true, performs non-blocking (hot) backups.	true	Optional: {}
jobTemplate BackupJobTemplate	JobTemplate is propagated to every generated Backup as its spec.jobTemplate, shaping the backup worker Job (resources, scheduling, labels/annotations, and the finished-Job TTL). When unset the generated Backups fall back to the cluster-wide spec.backup.jobTemplate.		Optional: {}
logical LogicalBackupOptions	Logical is propagated to every generated Backup as its spec.logical when method is "logical".		Optional: {}
successfulBackupsHistoryLimit integer	SuccessfulBackupsHistoryLimit caps how many completed Backup objects this schedule keeps. The newest that many are retained and older completed Backups are garbage-collected. Unset means no count limit. The single newest completed Backup is always kept regardless, so a schedule never prunes its last recovery point. Deleting a Backup honours its reclaimPolicy: a Delete-policy Backup also reclaims its object-store archive, a Retain-policy one leaves the archive (see spec.backup.retentionPolicy on the Cluster for object-store retention).		Minimum: 0 Optional: {}
failedBackupsHistoryLimit integer	FailedBackupsHistoryLimit caps how many failed Backup objects this schedule keeps. The newest that many are retained and older failed Backups are garbage-collected. Unset means no count limit.		Minimum: 0 Optional: {}
retentionPolicy string	RetentionPolicy is a time window (e.g. "30d", "8w", "3m"; days, weeks, months, where a month is 30 days) after which this schedule's terminal Backup objects are garbage-collected. It uses the same syntax as the Cluster spec.backup.retentionPolicy. A terminal Backup is pruned when it exceeds the history limit OR ages past this window, whichever applies. Unset means no time limit. The newest completed Backup is always kept.		Pattern: ^[1-9][0-9]*[dwm]$ Optional: {}

ScheduledBackupStatus

ScheduledBackupStatus defines the observed state of ScheduledBackup.

Appears in:

• ScheduledBackup

Field	Description	Default	Validation
lastCheckTime Time	LastCheckTime is the last time the schedule was evaluated.		Optional: {}
lastScheduleTime Time	LastScheduleTime is the last time a backup was triggered.		Optional: {}
nextScheduleTime Time	NextScheduleTime is the next time a backup will be triggered.		Optional: {}

SecretKeySelector

SecretKeySelector selects a key from a Secret in the same namespace.

Appears in:

• DatabaseUserSpec
• ExternalCluster
• InlineUser
• MonitoringConfiguration
• RoleConfiguration
• S3Credentials
• S3TLSConfig

Field	Description	Default	Validation
name string	Name of the Secret		Required: {}
key string	Key within the Secret to select		Required: {}

SemiSyncConfiguration

SemiSyncConfiguration configures semi-synchronous replication.

Appears in:

• MySQLConfiguration

Field	Description	Default	Validation
enabled boolean	Enabled turns on semi-synchronous replication.	false	Optional: {}
timeoutMillis integer	Timeout in milliseconds the primary waits for a replica acknowledgement before falling back to asynchronous replication.		Minimum: 0 Optional: {}
dataDurability string	DataDurability controls how strictly minSyncReplicas is enforced when replicas are unhealthy. "preferred" (the default) keeps the primary writable by self-healing the acknowledgement count down to the number of healthy replicas; "required" leaves it fixed so writes block until enough replicas acknowledge. It has no effect on MariaDB, whose primary always waits for exactly one acknowledgement.	preferred	Enum: [preferred required] Optional: {}

ServiceAccountTemplate

ServiceAccountTemplate contains the template needed to generate the service accounts.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
metadata Metadata	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}

ServiceSelectorType

Underlying type: string

ServiceSelectorType is the type of a default service.

Validation:

• Enum: [rw ro r]

Appears in:

• ManagedService
• ManagedServices

Field	Description
rw	ServiceSelectorTypeRW selects the read-write (primary) service.
ro	ServiceSelectorTypeRO selects the read-only (replicas) service.
r	ServiceSelectorTypeR selects the read (any instance) service.

ServiceTemplateServiceSpec

ServiceTemplateServiceSpec exposes the subset of corev1.ServiceSpec fields that users are allowed to customise. The operator retains control over the selector, ports, clusterIP, and owner reference.

Appears in:

• ServiceTemplateSpec

Field	Description	Default	Validation
type ServiceType	Type determines how the Service is exposed. Defaults to ClusterIP.		Enum: [ClusterIP NodePort LoadBalancer ExternalName] Optional: {}
externalTrafficPolicy ServiceExternalTrafficPolicy	ExternalTrafficPolicy describes how nodes distribute service traffic.		Optional: {}
sessionAffinity ServiceAffinity	SessionAffinity configures session affinity.		Optional: {}
loadBalancerSourceRanges string array	LoadBalancerSourceRanges restricts load balancer access.		Optional: {}
externalName string	ExternalName is the external reference for ExternalName services.		Optional: {}
healthCheckNodePort integer	HealthCheckNodePort specifies the health check node port.		Optional: {}

ServiceTemplateSpec

ServiceTemplateSpec describes the user-customisable parts of a managed Service.

Appears in:

• ManagedService
• ManagedServices

Field	Description	Default	Validation
metadata ObjectMetaTemplate	Refer to Kubernetes API documentation for fields of metadata.		Optional: {}
spec ServiceTemplateServiceSpec	Specification of the desired behavior of the Service. The selector field is operator-managed and cannot be overridden.		Optional: {}

ServiceUpdateStrategy

Underlying type: string

ServiceUpdateStrategy describes how the service template is reconciled.

Validation:

• Enum: [patch replace]

Appears in:

• ManagedService

Field	Description
patch	ServiceUpdateStrategyPatch merges user fields onto operator defaults.
replace	ServiceUpdateStrategyReplace replaces operator defaults with the user template.

StorageConfiguration

StorageConfiguration is the configuration used to create and reconcile PVCs, usable for the instance data directory or for separate binlog storage.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
storageClass string	StorageClass to use for PVCs. Applied after evaluating the PVC template, if available. If not specified, the generated PVCs will use the default storage class.		Optional: {}
size string	Size of the storage. Required if not already specified in the PVC template. Changes to this field are automatically reapplied to the created PVCs. Size cannot be decreased.		Optional: {}
resizeInUseVolumes boolean	ResizeInUseVolumes, when true (the default), grows mounted PVCs in place and relies on the storage backend to expand the filesystem online. Set it to false when the backend cannot expand a volume in use: the operator then completes a resize by recycling the instance Pod (serialised replica-by-replica, primary last) so the volume is detached and remounted.	true	Optional: {}
pvcTemplate PersistentVolumeClaimSpec	Template to be used to generate the Persistent Volume Claim		Optional: {}

UpgradeConfiguration

UpgradeConfiguration tunes MySQL server major-version upgrades.

Appears in:

• ClusterSpec

Field	Description	Default	Validation
backupBeforeUpgrade boolean	BackupBeforeUpgrade controls whether the operator takes a fresh backup before starting a major-version upgrade and waits for it to succeed before rolling any instance. Defaults to true. Set false to skip (e.g. when an external backup process is in place). The data-dictionary upgrade is irreversible, so the backup is the only rollback path.		Optional: {}