Skip to content

Latest commit

 

History

History
642 lines (464 loc) · 47.5 KB

cluster_spec.md

File metadata and controls

642 lines (464 loc) · 47.5 KB
Raw

API Docs

This Document documents the types introduced by the FoundationDB Operator to be consumed by users.

Note this document is generated from code comments. When contributing a change to this document please do so by changing the code comments.

Table of Contents

AutomaticReplacementOptions

AutomaticReplacementOptions controls options for automatically replacing failed processes.

Field Description Scheme Required
enabled Enabled controls whether automatic replacements are enabled. The default is false. *bool false
faultDomainBasedReplacements FaultDomainBasedReplacements controls whether automatic replacements are targeting all failed process groups in a fault domain or only specific Process Groups. If this setting is enabled, the number of different fault domains that can have all their failed process groups replaced at the same time will be equal to MaxConcurrentReplacements. e.g. MaxConcurrentReplacements = 2 would mean that at most 2 different fault domains can have their failed process groups replaced at the same time. The default is false. *bool false
failureDetectionTimeSeconds FailureDetectionTimeSeconds controls how long a process must be failed or missing before it is automatically replaced. The default is 7200 seconds, or 2 hours. *int false
taintReplacementTimeSeconds TaintReplacementTimeSeconds controls how long a pod stays in NodeTaintReplacing condition before it is automatically replaced. The default is 1800 seconds, i.e., 30min *int false
maxConcurrentReplacements MaxConcurrentReplacements controls how many automatic replacements are allowed to take part. This will take the list of current replacements and then calculate the difference between maxConcurrentReplacements and the size of the list. e.g. if currently 3 replacements are queued (e.g. in the processGroupsToRemove list) and maxConcurrentReplacements is 5 the operator is allowed to replace at most 2 process groups. Setting this to 0 will basically disable the automatic replacements. *int false
taintReplacementOptions TaintReplacementOption controls which taint label the operator will react to. []TaintReplacementOption false
maxFaultDomainsWithTaintedProcessGroups MaxFaultDomainsWithTaintedProcessGroups defines how many fault domains in the cluster can have process groups with the NodeTaintReplacing condition and still allow the operator to automatically replace those process groups. If more fault domains contain process groups with the NodeTaintReplacing condition, the operator will not automatically replace those process groups. This is a safeguard in addition to MaxConcurrentReplacements to make sure the operator is not replacing too many process groups if a large number of nodes are tainted. A absolute number of fault domains or a percentage can be provided. Defaults to 10% of the fault domains or at least 1. *intstr.IntOrString false

Back to TOC

BuggifyConfig

BuggifyConfig provides options for injecting faults into a cluster for testing.

Field Description Scheme Required
noSchedule NoSchedule defines a list of process group IDs that should fail to schedule. []ProcessGroupID false
crashLoop CrashLoops defines a list of process group IDs that should be put into a crash looping state. Deprecated: use CrashLoopContainers instead. []ProcessGroupID false
crashLoopContainers CrashLoopContainers defines a list of process group IDs and containers that should be put into a crash looping state. []CrashLoopContainerObject false
emptyMonitorConf EmptyMonitorConf instructs the operator to update all of the fdbmonitor.conf files to have zero fdbserver processes configured. bool false
ignoreDuringRestart IgnoreDuringRestart instructs the operator to ignore the provided process groups IDs during the restart command. This can be useful to simulate cases where the kill command is not restarting all processes. IgnoreDuringRestart does not support the wildcard option to ignore all of this specific cluster processes. []ProcessGroupID false
blockRemoval BlockRemoval defines a list of process group IDs that will not be removed, even if they are marked for removal. The operator will trigger the exclusion but the removal of the resources will be blocked until they are removed from this list. This setting can be used to simulate cases where a process group is marked for removal but the resources are not yet removed. []ProcessGroupID false

Back to TOC

ClusterGenerationStatus

ClusterGenerationStatus stores information on which generations have reached different stages in reconciliation for the cluster.

Field Description Scheme Required
reconciled Reconciled provides the last generation that was fully reconciled. int64 false
needsConfigurationChange NeedsConfigurationChange provides the last generation that is pending a change to configuration. int64 false
needsCoordinatorChange NeedsCoordinatorChange provides the last generation that is pending a change to its coordinators. int64 false
needsBounce NeedsBounce provides the last generation that is pending a bounce of fdbserver. int64 false
needsPodDeletion NeedsPodDeletion provides the last generation that is pending pods being deleted and recreated. int64 false
needsShrink NeedsShrink provides the last generation that is pending pods being excluded and removed. int64 false
needsGrow NeedsGrow provides the last generation that is pending pods being added. int64 false
needsMonitorConfUpdate NeedsMonitorConfUpdate provides the last generation that needs an update through the fdbmonitor conf. int64 false
missingDatabaseStatus DatabaseUnavailable provides the last generation that could not complete reconciliation due to the database being unavailable. int64 false
hasExtraListeners HasExtraListeners provides the last generation that could not complete reconciliation because it has more listeners than it is supposed to. int64 false
needsServiceUpdate NeedsServiceUpdate provides the last generation that needs an update to the service config. int64 false
hasPendingRemoval HasPendingRemoval provides the last generation that has pods that have been excluded but are pending being removed. A cluster in this state is considered reconciled, but we track this in the status to allow users of the operator to track when the removal is fully complete. int64 false
hasUnhealthyProcess HasUnhealthyProcess provides the last generation that has at least one process group with a negative condition. int64 false
needsLockConfigurationChanges NeedsLockConfigurationChanges provides the last generation that is pending a change to the configuration of the locking system. int64 false

Back to TOC

ClusterHealth

ClusterHealth represents different views into health in the cluster status.

Field Description Scheme Required
available Available reports whether the database is accepting reads and writes. bool false
healthy Healthy reports whether the database is in a fully healthy state. bool false
fullReplication FullReplication reports whether all data are fully replicated according to the current replication policy. bool false
dataMovementPriority DataMovementPriority reports the priority of the highest-priority data movement in the cluster. int false

Back to TOC

ConnectionString

ConnectionString models the contents of a cluster file in a structured way

Field Description Scheme Required
databaseName DatabaseName provides an identifier for the database which persists across coordinator changes. string false
generationID GenerationID provides a unique ID for the current generation of coordinators. string false
coordinators Coordinators provides the addresses of the current coordinators. []string false

Back to TOC

ContainerOverrides

ContainerOverrides provides options for customizing a container created by the operator.

Field Description Scheme Required
enableLivenessProbe EnableLivenessProbe defines if the sidecar should have a livenessProbe. This setting will be ignored on the main container. *bool false
enableReadinessProbe EnableReadinessProbe defines if the sidecar should have a readinessProbe. This setting will be ignored on the main container. Deprecated: Will be removed in the next major release. *bool false
enableTls EnableTLS controls whether we should be listening on a TLS connection. bool false
peerVerificationRules PeerVerificationRules provides the rules for what client certificates the process should accept. string false
imageConfigs ImageConfigs allows customizing the image that we use for a container. []ImageConfig false

Back to TOC

CoordinatorSelectionSetting

CoordinatorSelectionSetting defines the process class and the priority of it. A higher priority means that the process class is preferred over another.

Field Description Scheme Required
processClass ProcessClass defines the process class to associate with priority with. ProcessClass false
priority Priority defines the ordering of different process classes. int false

Back to TOC

CrashLoopContainerObject

CrashLoopContainerObject specifies crash-loop target for specific container.

Field Description Scheme Required
containerName Name of the target container. string false
targets Target processes to kill inside the container. []ProcessGroupID false

Back to TOC

FaultDomain

FaultDomain represents the FaultDomain of a process group

Back to TOC

FoundationDBCluster

FoundationDBCluster is the Schema for the foundationdbclusters API

Field Description Scheme Required
metadata metav1.ObjectMeta false
spec FoundationDBClusterSpec false
status FoundationDBClusterStatus false

Back to TOC

FoundationDBClusterAutomationOptions

FoundationDBClusterAutomationOptions provides flags for enabling or disabling operations that can be performed on a cluster.

Field Description Scheme Required
configureDatabase ConfigureDatabase defines whether the operator is allowed to reconfigure the database. *bool false
killProcesses KillProcesses defines whether the operator is allowed to bounce fdbserver processes. *bool false
cacheDatabaseStatusForReconciliation CacheDatabaseStatusForReconciliation defines whether the operator is using the same FoundationDB machine-readable status for all sub-reconcilers or if the machine-readable status should be fetched by ever sub-reconciler if required. Enabling this setting might improve the operator reconciliation speed for large clusters. *bool false
replacements Replacements contains options for automatically replacing failed processes. AutomaticReplacementOptions false
ignorePendingPodsDuration IgnorePendingPodsDuration defines how long a Pod has to be in the Pending Phase before ignore it during reconciliation. This prevents Pod that are stuck in Pending to block further reconciliation. time.Duration false
useNonBlockingExcludes UseNonBlockingExcludes defines whether the operator is allowed to use non blocking exclude commands. The default is false. *bool false
useLocalitiesForExclusion UseLocalitiesForExclusion defines whether the exclusions are done using localities instead of IP addresses. This feature requires at least FDB 7.1.42 or 7.3.26. The default is false. *bool false
ignoreTerminatingPodsSeconds IgnoreTerminatingPodsSeconds defines how long a Pod has to be in the Terminating Phase before we ignore it during reconciliation. This prevents Pod that are stuck in Terminating to block further reconciliation. *int false
ignoreMissingProcessesSeconds IgnoreMissingProcessesSeconds defines how long a process group has to be in the MissingProcess condition until it will be ignored during reconciliation. This prevents that a process will block reconciliation. *int false
failedPodDurationSeconds FailedPodDurationSeconds defines the duration a Pod can stay in the deleted state (deletionTimestamp != 0) before it gets marked as PodFailed. This is important in cases where a fdbserver process is still reporting but the Pod resource is marked for deletion. This can happen when the kubelet or a node fails. Setting this condition will ensure that the operator is replacing affected Pods. *int false
maxConcurrentReplacements MaxConcurrentReplacements defines how many process groups can be concurrently replaced if they are misconfigured. If the value will be set to 0 this will block replacements and these misconfigured Pods must be replaced manually or by another process. For each reconcile loop the operator calculates the maximum number of possible replacements by taken this value as the upper limit and removes all ongoing replacements that have not finished. Which means if the value is set to 5 and we have 4 ongoing replacements (process groups marked with remove but not excluded) the operator is allowed to replace on further process group. *int false
deletionMode DeletionMode defines the deletion mode for this cluster. This can be PodUpdateModeNone, PodUpdateModeAll, PodUpdateModeZone or PodUpdateModeProcessGroup. The DeletionMode defines how Pods are deleted in order to update them or when they are removed. PodUpdateMode false
removalMode RemovalMode defines the removal mode for this cluster. This can be PodUpdateModeNone, PodUpdateModeAll, PodUpdateModeZone or PodUpdateModeProcessGroup. The RemovalMode defines how process groups are deleted in order when they are marked for removal. PodUpdateMode false
waitBetweenRemovalsSeconds WaitBetweenRemovalsSeconds defines how long to wait between the last removal and the next removal. This is only an upper limit if the process group and the according resources are deleted faster than the provided duration the operator will move on with the next removal. The idea is to prevent a race condition were the operator deletes a resource but the Kubernetes API is slower to trigger the actual deletion, and we are running into a situation where the fault tolerance check still includes the already deleted processes. Defaults to 60. *int false
podUpdateStrategy PodUpdateStrategy defines how Pod spec changes are rolled out either by replacing Pods or by deleting Pods. The default for this is ReplaceTransactionSystem. PodUpdateStrategy false
useManagementAPI UseManagementAPI defines if the operator should make use of the management API instead of using fdbcli to interact with the FoundationDB cluster. *bool false
maintenanceModeOptions MaintenanceModeOptions contains options for maintenance mode related settings. MaintenanceModeOptions false
ignoreLogGroupsForUpgrade IgnoreLogGroupsForUpgrade defines the list of LogGroups that should be ignored during fdb version upgrade. The default is a list that includes "fdb-kubernetes-operator". []LogGroup false

Back to TOC

FoundationDBClusterFaultDomain

FoundationDBClusterFaultDomain describes the fault domain that a cluster is replicated across.

Field Description Scheme Required
key Key provides a topology key for the fault domain to replicate across. string false
value Value provides a harcoded value to use for the zoneid for the pods. string false
valueFrom ValueFrom provides a field selector to use as the source of the fault domain. string false
zoneCount ZoneCount provides the number of fault domains in the data center where these processes are running. This is only used in the kubernetes-cluster fault domain strategy. int false
zoneIndex ZoneIndex provides the index of this Kubernetes cluster in the list of KCs in the data center. This is only used in the kubernetes-cluster fault domain strategy. int false

Back to TOC

FoundationDBClusterList

FoundationDBClusterList contains a list of FoundationDBCluster objects

Field Description Scheme Required
metadata metav1.ListMeta false
items []FoundationDBCluster true

Back to TOC

FoundationDBClusterSpec

FoundationDBClusterSpec defines the desired state of a cluster.

Field Description Scheme Required
version Version defines the version of FoundationDB the cluster should run. string true
databaseConfiguration DatabaseConfiguration defines the database configuration. DatabaseConfiguration false
processes Processes defines process-level settings. map[ProcessClass]ProcessSettings false
processCounts ProcessCounts defines the number of processes to configure for each process class. You can generally omit this, to allow the operator to infer the process counts based on the database configuration. ProcessCounts false
seedConnectionString SeedConnectionString provides an additional connection string. This connection string will be used in addition to the connection string defined under cluster.Status.ConnectionString to connect to the cluster. This setting can be used to create a multi-region cluster or to recover a cluster if it is out of sync. string false
partialConnectionString PartialConnectionString provides a way to specify part of the connection string (e.g. the database name and coordinator generation) without specifying the entire string. This does not allow for setting the coordinator IPs. If SeedConnectionString is set, PartialConnectionString will have no effect. They cannot be used together. ConnectionString false
faultDomain FaultDomain defines the rules for what fault domain to replicate across. FoundationDBClusterFaultDomain false
processGroupsToRemove ProcessGroupsToRemove defines the process groups that we should remove from the cluster. This list contains the process group IDs. []ProcessGroupID false
processGroupsToRemoveWithoutExclusion ProcessGroupsToRemoveWithoutExclusion defines the process groups that we should remove from the cluster without excluding them. This list contains the process group IDs. This should be used for cases where a pod does not have an IP address and you want to remove it and destroy its volume without confirming the data is fully replicated. []ProcessGroupID false
configMap ConfigMap allows customizing the config map the operator creates. *corev1.ConfigMap false
mainContainer MainContainer defines customization for the foundationdb container. ContainerOverrides false
sidecarContainer SidecarContainer defines customization for the foundationdb-kubernetes-sidecar container. ContainerOverrides false
trustedCAs TrustedCAs defines a list of root CAs the cluster should trust, in PEM format. []string false
sidecarVariables SidecarVariables defines Custom variables that the sidecar should make available for substitution in the monitor conf file. []string false
logGroup LogGroup defines the log group to use for the trace logs for the cluster. string false
dataCenter DataCenter defines the data center where these processes are running. string false
dataHall DataHall defines the data hall where these processes are running. string false
automationOptions AutomationOptions defines customization for enabling or disabling certain operations in the operator. FoundationDBClusterAutomationOptions false
processGroupIDPrefix ProcessGroupIDPrefix defines a prefix to append to the process group IDs in the locality fields. This must be a valid Kubernetes label value. See https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#syntax-and-character-set for more details on that. string false
lockOptions LockOptions allows customizing how we manage locks for global operations. LockOptions false
routing Routing defines the configuration for routing to our pods. RoutingConfig false
ignoreUpgradabilityChecks IgnoreUpgradabilityChecks determines whether we should skip the check for client compatibility when performing an upgrade. bool false
buggify Buggify defines settings for injecting faults into a cluster for testing. BuggifyConfig false
storageServersPerPod StorageServersPerPod defines how many Storage Servers should run in a single process group (Pod). This number defines the number of processes running in one Pod whereas the ProcessCounts defines the number of Pods created. This means that you end up with ProcessCounts["storage"] * StorageServersPerPod storage processes. int false
logServersPerPod LogServersPerPod defines how many Log Servers should run in a single process group (Pod). This number defines the number of processes running in one Pod whereas the ProcessCounts defines the number of Pods created. This means that you end up with ProcessCounts["Log"] * LogServersPerPod log processes. This also affects processes with the transaction class. int false
minimumUptimeSecondsForBounce MinimumUptimeSecondsForBounce defines the minimum time, in seconds, that the processes in the cluster must have been up for before the operator can execute a bounce. int false
replaceInstancesWhenResourcesChange ReplaceInstancesWhenResourcesChange defines if an instance should be replaced when the resource requirements are increased. This can be useful with the combination of local storage. *bool false
skip Skip defines if the cluster should be skipped for reconciliation. This can be useful for investigating in issues or if the environment is unstable. bool false
coordinatorSelection CoordinatorSelection defines which process classes are eligible for coordinator selection. If empty all stateful processes classes are equally eligible. A higher priority means that a process class is preferred over another process class. If the FoundationDB cluster is spans across multiple Kubernetes clusters or DCs the CoordinatorSelection must match in all FoundationDB cluster resources otherwise the coordinator selection process could conflict. []CoordinatorSelectionSetting false
labels LabelConfig allows customizing labels used by the operator. LabelConfig false
useExplicitListenAddress UseExplicitListenAddress determines if we should add a listen address that is separate from the public address. Deprecated: This setting will be removed in the next major release. *bool false
imageType ImageType defines the image type that should be used for the FoundationDBCluster deployment. When the type is set to "unified" the deployment will use the new fdb-kubernetes-monitor. Otherwise the main container and the sidecar container will use different images. Default: split *ImageType false
maxZonesWithUnavailablePods MaxZonesWithUnavailablePods defines the maximum number of zones that can have unavailable pods during the update process. When unset, there is no limit to the number of zones with unavailable pods. *int false

Back to TOC

FoundationDBClusterStatus

FoundationDBClusterStatus defines the observed state of FoundationDBCluster

Field Description Scheme Required
databaseConfiguration DatabaseConfiguration provides the running configuration of the database. DatabaseConfiguration false
generations Generations provides information about the latest generation to be reconciled, or to reach other stages at which reconciliation can halt. ClusterGenerationStatus false
health Health provides information about the health of the database. ClusterHealth false
requiredAddresses RequiredAddresses define that addresses that we need to enable for the processes in the cluster. RequiredAddressSet false
hasIncorrectConfigMap HasIncorrectConfigMap indicates whether the latest config map is out of date with the cluster spec. bool false
hasIncorrectServiceConfig HasIncorrectServiceConfig indicates whether the cluster has service config that is out of date with the cluster spec. bool false
needsNewCoordinators NeedsNewCoordinators indicates whether the cluster needs to recruit new coordinators to fulfill its fault tolerance requirements. bool false
runningVersion RunningVersion defines the version of FoundationDB that the cluster is currently running. string false
connectionString ConnectionString defines the contents of the cluster file. string false
configured Configured defines whether we have configured the database yet. bool false
hasListenIPsForAllPods HasListenIPsForAllPods defines whether every pod has an environment variable for its listen address. bool false
storageServersPerDisk StorageServersPerDisk defines the storageServersPerPod observed in the cluster. If there are more than one value in the slice the reconcile phase is not finished. []int false
logServersPerDisk LogServersPerDisk defines the LogServersPerDisk observed in the cluster. If there are more than one value in the slice the reconcile phase is not finished. []int false
imageTypes ImageTypes defines the kinds of images that are in use in the cluster. If there is more than one value in the slice the reconcile phase is not finished. []ImageType false
processGroups ProcessGroups contain information about a process group. This information is used in multiple places to trigger the according action. []*ProcessGroupStatus false
locks Locks contains information about the locking system. LockSystemStatus false
maintenanceModeInfo MaintenenanceModeInfo contains information regarding process groups in maintenance mode Deprecated: This setting is not used anymore. MaintenanceModeInfo false
desiredProcessGroups DesiredProcessGroups reflects the number of expected running process groups. int false
reconciledProcessGroups ReconciledProcessGroups reflects the number of process groups that have no condition and are not marked for removal. int false

Back to TOC

ImageType

ImageType defines a single kind of images used in the cluster.

Back to TOC

LabelConfig

LabelConfig allows customizing labels used by the operator.

Field Description Scheme Required
matchLabels MatchLabels provides the labels that the operator should use to identify resources owned by the cluster. These will automatically be applied to all resources the operator creates. map[string]string false
resourceLabels ResourceLabels provides additional labels that the operator should apply to resources it creates. map[string]string false
processGroupIDLabels ProcessGroupIDLabels provides the labels that we use for the process group ID field. The first label will be used by the operator when filtering resources. []string false
processClassLabels ProcessClassLabels provides the labels that we use for the process class field. The first label will be used by the operator when filtering resources. []string false
filterOnOwnerReference FilterOnOwnerReferences determines whether we should check that resources are owned by the cluster object, in addition to the constraints provided by the match labels. Deprecated: This setting will be removed in the next major release. *bool false

Back to TOC

LockDenyListEntry

LockDenyListEntry models an entry in the deny list for the locking system.

Field Description Scheme Required
id The ID of the operator instance this entry is targeting. string false
allow Whether the instance is allowed to take locks. bool false

Back to TOC

LockOptions

LockOptions provides customization for locking global operations.

Field Description Scheme Required
disableLocks DisableLocks determines whether we should disable locking entirely. *bool false
lockKeyPrefix LockKeyPrefix provides a custom prefix for the keys in the database we use to store locks. string false
lockDurationMinutes LockDurationMinutes determines the duration that locks should be valid for. *int false
denyList DenyList manages configuration for whether an instance of the operator should be denied from taking locks. []LockDenyListEntry false

Back to TOC

LockSystemStatus

LockSystemStatus provides a summary of the status of the locking system.

Field Description Scheme Required
lockDenyList DenyList contains a list of operator instances that are prevented from taking locks. []string false

Back to TOC

LogGroup

LogGroup represents a LogGroup used by a FoundationDB process to log trace events. The LogGroup can be used to filter clients during an upgrade.

Back to TOC

MaintenanceModeInfo

MaintenanceModeInfo contains information regarding the zone and process groups that are put into maintenance mode by the operator

Field Description Scheme Required
startTimestamp StartTimestamp provides the timestamp when this zone is put into maintenance mode Deprecated: This setting is not used anymore. *metav1.Time false
zoneID ZoneID that is placed in maintenance mode Deprecated: This setting is not used anymore. FaultDomain false
processGroups ProcessGroups that are placed in maintenance mode Deprecated: This setting is not used anymore. []string false

Back to TOC

MaintenanceModeOptions

MaintenanceModeOptions controls options for placing zones in maintenance mode.

Field Description Scheme Required
UseMaintenanceModeChecker UseMaintenanceModeChecker defines whether the operator is allowed to use maintenance mode before updating pods. If this setting is set to true the operator will set and reset the maintenance mode when updating pods. If this setting is set to true, then ResetMaintenanceMode will also be enabled to make sure the operator is able to reset the maintenance mode again. Default is false. *bool false
resetMaintenanceMode ResetMaintenanceMode defines whether the operator should reset the maintenance mode if all storage processes under the maintenance zone have been restarted. The default is false. For more details see: https://github.com/FoundationDB/fdb-kubernetes-operator/blob/improve-maintenance-mode-integration/docs/manual/operations.md#maintenance Default is false. *bool false
maintenanceModeTimeSeconds MaintenanceModeTimeSeconds provides the duration for the zone to be in maintenance. It will automatically be switched off after the time elapses. Default is 600. *int false

Back to TOC

PodUpdateMode

PodUpdateMode defines the deletion mode for the cluster

Back to TOC

PodUpdateStrategy

PodUpdateStrategy defines how Pod spec changes should be applied.

Back to TOC

ProcessGroupCondition

ProcessGroupCondition represents a degraded condition that a process group is in.

Field Description Scheme Required
type Name of the condition ProcessGroupConditionType false
timestamp Timestamp when the Condition was observed int64 false

Back to TOC

ProcessGroupConditionType

ProcessGroupConditionType represents a concrete ProcessGroupCondition.

Back to TOC

ProcessGroupID

ProcessGroupID represents the ID of the process group

Back to TOC

ProcessGroupStatus

ProcessGroupStatus represents the status of a ProcessGroup.

Field Description Scheme Required
processGroupID ProcessGroupID represents the ID of the process group ProcessGroupID false
processClass ProcessClass represents the class the process group has. ProcessClass false
addresses Addresses represents the list of addresses the process group has been known to have. []string false
removalTimestamp RemoveTimestamp if not empty defines when the process group was marked for removal. *metav1.Time false
exclusionTimestamp ExclusionTimestamp defines when the process group has been fully excluded. This is only used within the reconciliation process, and should not be considered authoritative. *metav1.Time false
exclusionSkipped ExclusionSkipped determines if exclusion has been skipped for a process, which will allow the process group to be removed without exclusion. bool false
processGroupConditions ProcessGroupConditions represents a list of degraded conditions that the process group is in. []*ProcessGroupCondition false
faultDomain FaultDomain represents the last seen fault domain from the cluster status. This can be used if a Pod or process is not running and would be missing in the cluster status. FaultDomain false

Back to TOC

ProcessSettings

ProcessSettings defines process-level settings.

Field Description Scheme Required
podTemplate PodTemplate allows customizing the pod. If a container image with a tag is specified the operator will throw an error and stop processing the cluster. *corev1.PodTemplateSpec false
volumeClaimTemplate VolumeClaimTemplate allows customizing the persistent volume claim for the pod. This will be ignored by the operator for stateless processes. *corev1.PersistentVolumeClaim false
customParameters CustomParameters defines additional parameters to pass to the fdbserver process. Only parameters for the [fdbserver] section are supported. Parameters from the [general] and [fdbmonitor] section are not supported. For more Information see: https://apple.github.io/foundationdb/configuration.html#general-section FoundationDBCustomParameters false

Back to TOC

PublicIPSource

PublicIPSource models options for how a pod gets its public IP.

Back to TOC

RequiredAddressSet

RequiredAddressSet provides settings for which addresses we need to listen on.

Field Description Scheme Required
tls TLS defines whether we need to listen on a TLS address. bool false
nonTLS NonTLS defines whether we need to listen on a non-TLS address. bool false

Back to TOC

RoutingConfig

RoutingConfig allows configuring routing to our pods, and services that sit in front of them.

Field Description Scheme Required
headlessService Headless determines whether we want to run a headless service for the cluster. *bool false
publicIPSource PublicIPSource specifies what source a process should use to get its public IPs. This supports the values pod and service. *PublicIPSource false
podIPFamily PodIPFamily tells the pod which family of IP addresses to use. You can use 4 to represent IPv4, and 6 to represent IPv6. This feature is only supported in FDB 7.0 or later, and requires dual-stack support in your Kubernetes environment. *int false
useDNSInClusterFile UseDNSInClusterFile determines whether to use DNS names rather than IP addresses to identify coordinators in the cluster file. This requires FoundationDB 7.0+. *bool false
defineDNSLocalityFields DefineDNSLocalityFields determines whether to define pod DNS names on pod specs and provide them in the locality arguments to fdbserver. This is ignored if UseDNSInCluster is true. *bool false
dnsDomain DNSDomain defines the cluster domain used in a DNS name generated for a service. The default is cluster.local. *string false

Back to TOC

TaintReplacementOption

TaintReplacementOption defines the taint key and taint duration the operator will react to a tainted node Example of TaintReplacementOption - key: "example.org/maintenance" durationInSeconds: 7200 # Ensure the taint is present for at least 2 hours before replacing Pods on a node with this taint. - key: "*" # The wildcard would allow to define a catch all configuration durationInSeconds: 3600 # Ensure the taint is present for at least 1 hour before replacing Pods on a node with this taint Setting durationInSeconds to the maximum of int64 will practically disable the taint key. When a Node taint key matches both an exact TaintReplacementOption key and a wildcard key, the exact matched key will be used.

Field Description Scheme Required
key Tainted key *string false
durationInSeconds The tainted key must be present for DurationInSeconds before operator replaces pods on the node with this taint; DurationInSeconds cannot be a negative number. *int64 false

Back to TOC

FoundationDBCustomParameter

FoundationDBCustomParameter defines a single custom knob

Back to TOC

DataCenter

DataCenter represents a data center in the region configuration

Field Description Scheme Required
id The ID of the data center. This must match the dcid locality field. string false
priority The priority of this data center when we have to choose a location. Higher priorities are preferred over lower priorities. int false
satellite Satellite indicates whether the data center is serving as a satellite for the region. A value of 1 indicates that it is a satellite, and a value of 0 indicates that it is not a satellite. int false

Back to TOC

DatabaseConfiguration

DatabaseConfiguration represents the configuration of the database

Field Description Scheme Required
redundancy_mode RedundancyMode defines the core replication factor for the database. RedundancyMode false
storage_engine StorageEngine defines the storage engine the database uses. StorageEngine false
usable_regions UsableRegions defines how many regions the database should store data in. int false
regions Regions defines the regions that the database can replicate in. []Region false
excluded_servers ExcludedServers defines the list of excluded servers form the database. []ExcludedServers false
RoleCounts RoleCounts defines how many processes the database should recruit for each role. RoleCounts true
VersionFlags VersionFlags defines internal flags for testing new features in the database. VersionFlags true
storage_migration_type StorageMigrationType defines the storage migration type. *StorageMigrationType false
perpetual_storage_wiggle PerpetualStorageWiggle defines the wiggle speed. If set to 0 this feature is disabled. When setting StorageMigrationType to StorageMigrationTypeGradual, this value must be greater than 0. *int false
perpetual_storage_wiggle_locality PerpetualStorageWiggleLocality if defined the specified locality will be migrated. Format is: <<LOCALITY_KEY>:<LOCALITY_VALUE>|0> *string false
perpetual_storage_wiggle_engine PerpetualStorageWiggleEngine defines the perpetual storage engine type. *StorageEngine false

Back to TOC

ExcludedServers

ExcludedServers represents the excluded servers in the database configuration

Field Description Scheme Required
address The Address of the excluded server. string false
locality The Locality of the excluded server. string false

Back to TOC

ProcessCounts

ProcessCounts represents the number of processes we have for each valid process class. If one of the counts in the spec is set to 0, we will infer the process count for that class from the role counts. If one of the counts in the spec is set to -1, we will not create any processes for that class. See GetProcessCountsWithDefaults for more information on the rules for inferring process counts.

Field Description Scheme Required
unset int false
storage int false
transaction int false
resolution int false
tester Deprecated: This setting will be removed in the next major release. use Test int false
test int false
proxy int false
commit_proxy int false
grv_proxy int false
master int false
stateless int false
log int false
cluster_controller int false
router int false
fast_restore int false
data_distributor int false
coordinator int false
ratekeeper int false
storage_cache int false
backup int false

Back to TOC

RedundancyMode

RedundancyMode defines the core replication factor for the database

Back to TOC

Region

Region represents a region in the database configuration

Field Description Scheme Required
datacenters The data centers in this region. []DataCenter false
satellite_logs The number of satellite logs that we should recruit. int false
satellite_redundancy_mode The replication strategy for satellite logs. RedundancyMode false

Back to TOC

RoleCounts

RoleCounts represents the roles whose counts can be customized.

Field Description Scheme Required
storage int false
logs int false
proxies int false
commit_proxies int false
grv_proxies int false
resolvers int false
log_routers int false
remote_logs int false

Back to TOC

StorageEngine

StorageEngine defines the storage engine for the database

Back to TOC

StorageMigrationType

StorageMigrationType defines the storage migration type.

Back to TOC

VersionFlags

VersionFlags defines internal flags for new features in the database.

Field Description Scheme Required
log_spill int false
log_version int false

Back to TOC

ProcessClass

ProcessClass models the class of a pod

Back to TOC

ImageConfig

ImageConfig provides a policy for customizing an image. When multiple image configs are provided, they will be merged into a single config that will be used to define the final image. For each field, we select the value from the first entry in the config list that defines a value for that field, and matches the version of FoundationDB the image is for. Any config that specifies a different version than the one under consideration will be ignored for the purposes of defining that image.

Field Description Scheme Required
version Version is the version of FoundationDB this policy applies to. If this is blank, the policy applies to all FDB versions. string false
baseImage BaseImage specifies the part of the image before the tag. string false
tag Tag specifies a full image tag. string false
tagSuffix TagSuffix specifies a suffix that will be added after the version to form the full tag. string false

Back to TOC