WekaContainer¶

kubectl explain wekacontainer

Field	Description	Default	Validation
`apiVersion` string	`weka.weka.io/v1alpha1`
`kind` string	`WekaContainer`
`metadata` ObjectMeta	Refer to Kubernetes API documentation for fields of `metadata`.
`spec` WekaContainerSpec
`status` WekaContainerStatus

WekaContainerSpec¶

kubectl explain wekacontainer.spec

Appears in: - WekaContainer

Field	Description	Default	Validation
`nodeAffinity` NodeName	name of the node where the container should run on
`failureDomain` FailureDomain	failure domain configuration
`topologySpreadConstraints` TopologySpreadConstraint array	controls the distribution of weka containers across the failure domains
`affinity` Affinity	advanced scheduling constraints
`nodeSelector` object (keys:string, values:string)
`port` integer
`exposePorts` integer array	deprecated, use ExposedPorts instead
`exposedPorts` ContainerPort array	ports to be exposed on the container, proxied to pod
`agentPort` integer
`portRange` PortRange
`image` string
`podConfigHash` string	a hash that represents the config state of a WekaContainer, will recreate the pod if stale
`imagePullSecret` string
`name` string
`mode` string			Enum: [drive compute client dist drivers-dist drivers-loader drivers-builder discovery s3 adhoc-op-with-container adhoc-op envoy nfs smbw telemetry ssdproxy data-services data-services-fe]
`numCores` integer
`extraCores` integer
`coreIds` integer array
`nonDatapathCoreIds` integer array	NonDatapathCoreIds pins management/aux (non-IONode) processes to specific CPUs.
`cpuPolicy` CpuPolicy		auto	Enum: [auto shared dedicated dedicated_ht manual]
`network` Network
`hugepages` integer
`hugepagesOffset` integer
`hugepagesSize` string
`numDrives` integer
`driversDistService` string
`driversLoaderImage` string
`driversBuildId` string
`builder` WekaContainerBuilder
`wekaSecretRef` EnvVarSource
`joinIpPorts` string array
`tracesConfiguration` TracesConfiguration
`tolerations` Toleration array
`nodeInfoConfigMap` string
`ipv6` boolean
`additionalMemory` integer
`group` string
`serviceAccountName` string
`additionalSecrets` object (keys:string, values:string)
`instructions` Instructions
`dropAffinityConstraints` boolean
`uploadResultsTo` string
`upgradePolicyType` UpgradePolicyType		manual	Enum: [manual all-at-once rolling all-at-once-force]
`state` ContainerState		active	Enum: [active paused destroying deleting]
`allowHotUpgrade` boolean
`driveCapacity` integer	DriveCapacity specifies the capacity (in GiB) per virtual drive, indicates this container uses shared drives via SSD proxy. When enabled, the container will: - Use virtual UUIDs instead of device paths for drives - Allocate capacity from shared drives rather than exclusive drives - Require an SSD proxy container to be running on the same node This value is copied from the cluster's DriveSharing.DriveCapacity configuration. Used to calculate total capacity request: NumDrives * DriveCapacity
`containerCapacity` integer	ContainerCapacity specifies the total capacity (in GiB) requested by this container when using shared drives via SSD proxy. This value takes precedence over DriveCapacity when both are set. It allows more flexible capacity allocation.
`driveTypesRatio` DriveTypesRatio	DriveTypesRatio specifies the desired ratio of drive types (TLC vs QLC) when allocating drives for the container.
`autoRemoveTimeout` Duration	sets weka cluster-side timeout, if client is not coming back in specified duration it will be auto removed from cluster config	0s	Pattern: `^(0\\|([0-9]+(\.[0-9]+)?(ns\\|us\\|µs\\|ms\\|s\\|m\\|h))+)$` Type: string
`overrides` WekaContainerSpecOverrides
`hostPID` boolean
`resources` PodResourcesSpec	resources to be proxied as-is to the pod spec
`pvc` PVCConfig
`dpdkBaseMemoryMb` integer
`dataServicesConfig` DataServicesConfig
`numa` WekaNuma	Numa configures NUMA confinement for this container

ContainerState¶

Underlying type: string

Appears in: - WekaContainerSpec

Field	Description
`active`
`paused`	Paused state indicates that the cluster which owns the container is in grace period before being destroyed
`destroying`	Destroying state indicates that the container can be destroyed without any additional steps (such as deactivation)
`deleting`	Deleting state indicates that the container is being deleted, normally such container should go through deactivation flow

CpuPolicy¶

Underlying type: string

Appears in: - WekaClientSpec - WekaClusterSpec - WekaContainerSpec

Field	Description
`auto`
`shared`
`dedicated`
`dedicated_ht`
`manual`

DataServicesConfig¶

kubectl explain wekacontainer.spec.dataServicesConfig

Appears in: - WekaContainerSpec

Field	Description	Default	Validation
`dataServicesFeCores` integer

DriveTypesRatio¶

kubectl explain wekacontainer.spec.driveTypesRatio

Appears in: - WekaClusterTemplate - WekaContainerSpec

Field	Description	Default	Validation
`tlc` integer		0	Minimum: 0
`qlc` integer		0	Minimum: 0

FailureDomain¶

kubectl explain wekacontainer.spec.failureDomain

Appears in: - WekaClusterSpec - WekaContainerSpec

Field	Description	Default	Validation
`label` string	label used for spreading the weka containers across different failure domains (if set) nodes that have the same value for the label will be considered as a single failure domain
`skew` integer	skew for the failure domain, if set, the weka containers will be spread with the skew in mind (only applicable if `label` is set)
`compositeLabels` string array	If multiple labels are specified, the failure domain will be the combination of the labels. If `compositeLabels` is set, `label` and `skew` will be ignored. When using compositeLabels, weka containers will be spread considering all labels with best effort, but even distribution is not guaranteed

Instructions¶

kubectl explain wekacontainer.spec.instructions

Appears in: - WekaContainerSpec

Field	Description	Default	Validation
`type` InstructionType
`payload` string

InstructionType¶

Underlying type: string

Appears in: - Instructions

Field	Description
`load-drivers`	Internal instructions — lifecycle plumbing, not exposed via any CRD.
`copy-weka-files-to-driver-loader`
`feature-flags-update`
`umount`

Network¶

kubectl explain wekacontainer.spec.network

Appears in: - RoleNetworkSelector - WekaClientSpec - WekaClusterSpec - WekaContainerSpec

Field	Description	Validation
`ethDevice` string	The name of a single network interface (for example, eth1) to be used by every backend container. This is for clusters that use only one dedicated NIC for the data path. You cannot use this field with ethDevices. If you leave this empty, the system automatically uses the node’s interface associated with the first subnet defined in deviceSubnets.
`ethDevices` string array	A list of network interface names to be used by backend containers when you have multiple dedicated NICs. The order of interfaces in this list is important, as it maps directly to the ethSlots index (the first interface maps to slot-0, the second to slot-1, and so on). You cannot use this field with ethDevice. Ensure that every interface listed here exists on all nodes that are part of the cluster.
`gateway` string	The default gateway IPv4 address for the backend containers’ data-path network. This is only necessary if backend subnets need to communicate with destinations outside of their local network (L2 segment). If you have a flat, non-routed backend network, you can leave this field empty.
`udpMode` boolean	A setting that enables or disables UDP encapsulation for backend traffic. - false (default): Uses standard raw Ethernet frames. true: Wraps data-path traffic in UDP packets. This is required if your network infrastructure or CNI (Container Network Interface) blocks traffic that isn’t IP-based.
`deviceSubnets` string array	A list of backend subnets in CIDR notation (for example, 192.168.10.0/24). The operator assigns IP addresses from these subnets to the backend containers for their data path network	items:Pattern: `^([0-9]\{1,3\}\.)\{3\}[0-9]\{1,3\}\/[0-9]\{1,2\}$`
`selectors` NetworkSelector array	Selectors define how backend data-path network interfaces are chosen on each node.
`managementIpsSelectors` NetworkSelector array	Selectors for management IPs used for cluster API and agent communication.
`bindManagementAll` boolean	BindManagementAll controls whether Weka containers bind to all network interfaces or only to specific management interfaces. When set to false (default), containers will only listen on the management ips interfaces (restrict_listen mode). When set to true, containers will listen on all ips (0.0.0.0) instead of specific IP addresses.
`nvidiaVfSingleIp` boolean	NvidiaVfSingleIp indicates whether NVIDIA virtual functions (VFs) should be configured to use a single-ip weka mode, where multiple weka processes can share same VF When not set defaults to false, in future releases, when auto-discovery of capabilities will be implemented not set might translate to true on supported setups
`allocateVfPerIoNode` boolean

NetworkSelector¶

kubectl explain wekacontainer.spec.network.selectors

Appears in: - Network

Field	Description	Default	Validation
`subnet` string	CIDR subnet (e.g. 192.168.10.0/24) to filter interfaces. Only interfaces with an IP in this subnet are eligible.
`min` integer	Minimum number of interfaces required from nodes matching this selector.
`max` integer	Maximum number of interfaces to select per node matching this selector.
`deviceNames` string array
`rdmaOnly` boolean
`disableRdma` boolean

NodeName¶

Underlying type: NodeName

Appears in: - ForceResignDrivesPayload - WekaContainerSpec - WekaContainerStatus

PVCConfig¶

kubectl explain wekacontainer.spec.pvc

Appears in: - WekaClientSpec - WekaClusterSpec - WekaContainerSpec

Field	Description	Default	Validation
`name` string	Name of the PersistentVolumeClaim to mount into all Weka containers.
`path` string	Mount path inside the Weka container. Defaults to /opt/k8s-weka when empty.

PodResourcesSpec¶

kubectl explain wekacontainer.spec.resources

Appears in: - DriverDistPayload - WekaClientSpec - WekaContainerSpec

Field	Description	Default	Validation
`requests` PodResources
`limits` PodResources

PodResources¶

kubectl explain wekacontainer.spec.resources.requests

Appears in: - PodResourcesSpec

Field	Description	Default	Validation
`cpu` Quantity
`memory` Quantity
`hugepages-2Mi` Quantity	Hugepages2Mi is requested verbatim as the pod's hugepages-2Mi resource. 1Gi pages are not settable through this field.

PortRange¶

kubectl explain wekacontainer.spec.portRange

Appears in: - WekaClientSpec - WekaContainerSpec

Field	Description	Default	Validation
`basePort` integer		45000	Maximum: 65535
`portRange` integer	number of ports to check for availability if 0 - will check all ports from basePort to 65535	0

TracesConfiguration¶

kubectl explain wekacontainer.spec.tracesConfiguration

TraceConfiguration defines the configuration for the traces, accepts parameters in gigabytes

Appears in: - WekaClientSpec - WekaClusterSpec - WekaContainerSpec

Field	Default	Validation
`maxCapacityPerIoNode` integer	10
`ensureFreeSpace` integer	20
`dumperConfigMode` DumperConfigMode	auto	Enum: [override partial-override auto cluster]

DumperConfigMode¶

Underlying type: string

Appears in: - TracesConfiguration

Field	Description
`auto`
`override`
`partial-override`
`cluster`

UpgradePolicyType¶

Underlying type: string

Appears in: - UpgradePolicy - WekaContainerSpec

Field	Description
`manual`
`rolling`
`all-at-once`
`all-at-once-force`

WekaContainerBuilder¶

Appears in: - WekaContainerSpec

Field	Description	Default	Validation
`weka_version` string

WekaContainerSpecOverrides¶

kubectl explain wekacontainer.spec.overrides

Appears in: - WekaContainerSpec

Field	Description	Default	Validation
`skipDeactivate` boolean	skips deactivation of container, this is unsafe operation that should be used only when this container will never be back into cluster
`skipDrivesForceResign` boolean	skips resign of drives, if we did not resign drives on removal of drive container we will not be able to reuse them, and manual operation with force resign will be required
`skipVirtualDrivesRemoval` boolean	skips removal of virtual drives from ssdproxy - unsafe operation that can lead to virtual drives leftovers should not be used unless instructed explicitly by weka personnel
`skipCleanupPersistentDir` boolean	skips cleanup of persistent directory, if this operation was omit local data of container will remain in persistent location(/opt/k8s-weka on vanilla OS/k8s distributions)
`upgradeForceReplace` boolean	unsafe operation, skips graceful stop of weka container for a quick replacement to a new image, should not be used unless instructed explicitly by weka personnel
`upgradePreventEviction` boolean
`podDeleteForceReplace` boolean
`machineIdentifierNodeRef` string
`preRunScript` string	script to be executed post initial persistency(if needed) configuration, before running actual workload
`forceDrain` boolean	unsafe operation, forces drain on the node where the container is running, should not be used unless instructed explicitly by weka personnel, the effect of drain is throwing away all IOs and acknowledging all umounts in unsafe manner
`skipActiveMountsCheck` boolean	option to skip active mounts check before deleting client containers
`umountOnHost` boolean	unsafe operation, runs nsenter in root namespace to umount all wekafs mounts visible on host
`debugSleepOnTerminate` integer	DebugSleepOnTerminate specifies the number of seconds to sleep on container abnormal exit for debugging purposes
`migrateOutFromPvc` boolean	MigrateOutFromPvc specifies that the container should be migrated out from PVC into local storage, this will be done prior to starting pod

WekaNuma¶

kubectl explain wekacontainer.spec.numa

WekaNuma configures NUMA confinement for a single weka container

Appears in: - RoleNumaSelector - WekaClientSpec - WekaClusterSpec - WekaContainerSpec

Field	Description	Validation
`single` boolean	Single, when true, confines the container to a single NUMA region	Optional: {}
`region` integer	Region is the NUMA region index to pin this container to	Minimum: 0 Optional: {}
`method` WekaNumaMethod	Method selects the enforcement mechanism	Enum: [device-plugin dra] Optional: {}

WekaNumaMethod¶

Underlying type: string

WekaNumaMethod defines how NUMA confinement is enforced

Validation: - Enum: [device-plugin dra]

Appears in: - WekaNuma

Field	Description
`device-plugin`
`dra`	WekaNumaMethodDra requests the NUMA region through a DRA ResourceClaim (driver numa.weka.io) instead of a device-plugin extended resource.

WekaContainerStatus¶

kubectl explain wekacontainer.status

Appears in: - WekaContainer

Field	Description	Default	Validation
`status` ContainerStatus		Init
`internalStatus` string
`managementIP` string
`managementIPs` string array
`containerID` integer
`clusterID` string
`conditions` Condition array
`lastAppliedImage` string
`lastAppliedSpec` string
`lastAppliedPodConfigHash` string
`nodeAffinity` NodeName
`result` string
`allocations` ContainerAllocations
`addedDrives` Drive array
`stats` WekaContainerMetrics
`printer` ContainerPrinterColumns
`timestamps` object (keys:string, values:Time)
`notToleratedOnReschedule` boolean

ContainerAllocations¶

Appears in: - WekaContainerStatus

Field	Description	Default	Validation
`drives` string array
`wekaPort` integer
`agentPort` integer
`failureDomain` string	value of the failure domain label of the node where the container is running
`machineIdentifier` string
`netDevices` string array
`virtualDrives` VirtualDrive array	VirtualDrives contains virtual drive allocations for drive sharing mode. Each VirtualDrive maps a virtual UUID to a physical drive UUID with allocated capacity.

VirtualDrive¶

VirtualDrive represents a virtual drive allocation in drive sharing mode

Appears in: - ContainerAllocations

Field	Description	Default	Validation
`virtualUUID` string	VirtualUUID is the virtual drive identifier
`physicalUUID` string	PhysicalUUID is the physical drive UUID obtained from proxy signing
`capacityGiB` integer	CapacityGiB is the allocated capacity in GiB
`serial` string	Serial is the serial number of the physical drive
`type` string	Type is the type of the drive (e.g., TLC, QLC)

ContainerPrinterColumns¶

Appears in: - WekaContainerStatus

Field	Description	Default	Validation
`processes` StringMetric
`drives` StringMetric
`activeMounts` StringMetric
`managementIPs` string	pretty-printed management IPs
`nodeAffinity` string	node name where the container is running
`capacity` string	per-drive-type capacity of a drive container, e.g. "TLC 30TiB / QLC 60TiB" (drive-sharing only)

StringMetric¶

Underlying type: string

Appears in: - ClientPrinterColumns - ClusterMetrics - ClusterPrinterColumns - ContainerPrinterColumns

ContainerStatus¶

Underlying type: string

Appears in: - WekaContainerStatus

Field	Description
`Init`
`PodNotRunning`
`PodRunning`
`PodTerminating`
`WaitForDrivers`
`Running`
`Stopped`
`Starting`
`Deleting`
`Destroying`
`Paused`
`Degraded`
`Unhealthy`
`Error`
`DrivesAdding`
`Draining`
`Completed`	for drivers-build and adhoc-op-with-container (sign-dives) container
`Building`
`StoppingAttempt`
`IoProcessesUp`	TimestampIoProcessesUp anchors the time IO processes were first observed up, for waitSinceIoProcessesUpTimeout.
`Stale`	Stale indicates the container's node has been removed from the k8s cluster and it is within the graceful-removal grace period before the operator deletes it.

Drive¶

Appears in: - WekaContainerStatus

Field	Description	Default	Validation
`uuid` string
`added_time` string
`device_path` string
`serial_number` string
`size_bytes` integer
`status` string

WekaContainerMetrics¶

Appears in: - WekaContainerStatus

Field	Description	Default	Validation
`processes` EntityStatefulNum
`cpuUtilization` FloatMetric
`drives` DriveMetrics
`activeMounts` IntMetric
`lastUpdate` Time

DriveMetrics¶

Appears in: - ClusterMetrics - WekaContainerMetrics

Field	Description	Default	Validation
`counters` EntityStatefulNum
`failures` DriveFailures array

DriveFailures¶

Appears in: - DriveMetrics

Field	Description	Default	Validation
`serialId` string
`wekaDriveId` string

EntityStatefulNum¶

Appears in: - ClientMetrics - ContainerMetrics - DriveMetrics - WekaContainerMetrics

Field	Description	Default	Validation
`active` IntMetric
`created` IntMetric
`desired` IntMetric

IntMetric¶

Underlying type: integer

Appears in: - CapacityMetrics - ClusterMetrics - EntityStatefulNum - FilesystemMetrics - StatusIops - StatusThroughput - WekaContainerMetrics

FloatMetric¶

Underlying type: string

Appears in: - ClusterMetrics - ContainerMetrics - MinMaxAvgPercent - WekaContainerMetrics