Documentation ¶
Index ¶
- Constants
- Variables
- type AllocBroadcaster
- type AllocExecRequest
- type AllocFileInfo
- type AllocHookResources
- type AllocListener
- type AllocResourceUsage
- type AllocStatsRequest
- type AllocStatsResponse
- type CSIControllerQuery
- type CSIVolumeMountOptions
- type ClientCSIControllerAttachVolumeRequest
- type ClientCSIControllerAttachVolumeResponse
- type ClientCSIControllerDetachVolumeRequest
- type ClientCSIControllerDetachVolumeResponse
- type ClientCSIControllerValidateVolumeRequest
- type ClientCSIControllerValidateVolumeResponse
- type ClientCSINodeDetachVolumeRequest
- type ClientCSINodeDetachVolumeResponse
- type ClientStatsResponse
- type CpuStats
- type FsListRequest
- type FsListResponse
- type FsLogsRequest
- type FsStatRequest
- type FsStatResponse
- type FsStreamRequest
- type HealthCheckIntervalRequest
- type HealthCheckIntervalResponse
- type HealthCheckRequest
- type HealthCheckResponse
- type MemoryStats
- type MonitorRequest
- type ResourceUsage
- type RpcError
- type StreamErrWrapper
- type TaskResourceUsage
Constants ¶
const CheckBufSize = 4 * 1024
CheckBufSize is the size of the buffer that is used for job output
Variables ¶
var DriverStatsNotImplemented = errors.New("stats not implemented for driver")
DriverStatsNotImplemented is the error to be returned if a driver doesn't implement stats.
var ErrAllocBroadcasterClosed = errors.New("alloc broadcaster closed")
Functions ¶
This section is empty.
Types ¶
type AllocBroadcaster ¶ added in v0.6.0
type AllocBroadcaster struct {
// contains filtered or unexported fields
}
AllocBroadcaster implements an allocation broadcast channel where each listener receives allocation updates. Pending updates are dropped and replaced by newer allocation updates, so listeners may not receive every allocation update. However this ensures Sends never block and listeners only receive the latest allocation update -- never a stale version.
func NewAllocBroadcaster ¶ added in v0.6.0
func NewAllocBroadcaster(l hclog.Logger) *AllocBroadcaster
NewAllocBroadcaster returns a new AllocBroadcaster.
func (*AllocBroadcaster) Close ¶ added in v0.6.0
func (b *AllocBroadcaster) Close()
Close closes the channel, disabling the sending of further allocation updates. Pending updates are still received by listeners. Safe to call concurrently and more than once.
func (*AllocBroadcaster) Listen ¶ added in v0.6.0
func (b *AllocBroadcaster) Listen() *AllocListener
Listen returns a Listener for the broadcast channel. New listeners receive the last sent alloc update.
func (*AllocBroadcaster) Send ¶ added in v0.6.0
func (b *AllocBroadcaster) Send(v *structs.Allocation) error
Send broadcasts an allocation update. Any pending updates are replaced with this version of the allocation to prevent blocking on slow receivers. Returns ErrAllocBroadcasterClosed if called after broadcaster is closed.
type AllocExecRequest ¶ added in v0.9.2
type AllocExecRequest struct { // AllocID is the allocation to stream logs from AllocID string // Task is the task to stream logs from Task string // Tty indicates whether to allocate a pseudo-TTY Tty bool // Cmd is the command to be executed Cmd []string structs.QueryOptions }
AllocExecRequest is the initial request for execing into an Alloc task
type AllocFileInfo ¶ added in v0.8.0
type AllocFileInfo struct { Name string IsDir bool Size int64 FileMode string ModTime time.Time ContentType string `json:",omitempty"` }
AllocFileInfo holds information about a file inside the AllocDir
type AllocHookResources ¶ added in v0.11.0
type AllocHookResources struct { CSIMounts map[string]*csimanager.MountInfo // contains filtered or unexported fields }
AllocHookResources contains data that is provided by AllocRunner Hooks for consumption by TaskRunners
func (*AllocHookResources) GetCSIMounts ¶ added in v0.11.0
func (a *AllocHookResources) GetCSIMounts() map[string]*csimanager.MountInfo
func (*AllocHookResources) SetCSIMounts ¶ added in v0.11.0
func (a *AllocHookResources) SetCSIMounts(m map[string]*csimanager.MountInfo)
type AllocListener ¶ added in v0.6.0
type AllocListener struct {
// contains filtered or unexported fields
}
AllocListener implements a listening endpoint for an allocation broadcast channel.
func (*AllocListener) Ch ¶ added in v0.6.0
func (l *AllocListener) Ch() <-chan *structs.Allocation
func (*AllocListener) Close ¶ added in v0.6.0
func (l *AllocListener) Close()
Close closes the Listener, disabling the receival of further messages. Safe to call more than once and concurrently with receiving on Ch.
type AllocResourceUsage ¶
type AllocResourceUsage struct { // ResourceUsage is the summation of the task resources ResourceUsage *ResourceUsage // Tasks contains the resource usage of each task Tasks map[string]*TaskResourceUsage // The max timestamp of all the Tasks Timestamp int64 }
AllocResourceUsage holds the aggregated task resource usage of the allocation.
type AllocStatsRequest ¶ added in v0.8.0
type AllocStatsRequest struct { // AllocID is the allocation to retrieves stats for AllocID string // Task is an optional filter to only request stats for the task. Task string structs.QueryOptions }
AllocStatsRequest is used to request the resource usage of a given allocation, potentially filtering by task
type AllocStatsResponse ¶ added in v0.8.0
type AllocStatsResponse struct { Stats *AllocResourceUsage structs.QueryMeta }
AllocStatsResponse is used to return the resource usage of a given allocation.
type CSIControllerQuery ¶ added in v0.11.0
type CSIControllerQuery struct { // ControllerNodeID is the node that should be targeted by the request ControllerNodeID string // PluginID is the plugin that should be targeted on the given node. PluginID string }
CSIControllerQuery is used to specify various flags for queries against CSI Controllers
type CSIVolumeMountOptions ¶ added in v0.11.0
type CSIVolumeMountOptions struct { // Filesystem is the desired filesystem type that should be used by the volume // (e.g ext4, aufs, zfs). This field is optional. Filesystem string // MountFlags contain the mount options that should be used for the volume. // These may contain _sensitive_ data and should not be leaked to logs or // returned in debugging data. // The total size of this field must be under 4KiB. MountFlags []string }
CSIVolumeMountOptions contains the mount options that should be provided when attaching and mounting a volume with the CSIVolumeAttachmentModeFilesystem attachment mode.
type ClientCSIControllerAttachVolumeRequest ¶ added in v0.11.0
type ClientCSIControllerAttachVolumeRequest struct { // The external ID of the volume to be used on a node. // This field is REQUIRED. VolumeID string // The ID of the node. This field is REQUIRED. This must match the NodeID that // is fingerprinted by the target node for this plugin name. ClientCSINodeID string // AttachmentMode indicates how the volume should be attached and mounted into // a task. AttachmentMode structs.CSIVolumeAttachmentMode // AccessMode indicates the desired concurrent access model for the volume AccessMode structs.CSIVolumeAccessMode // MountOptions is an optional field that contains additional configuration // when providing an AttachmentMode of CSIVolumeAttachmentModeFilesystem MountOptions *CSIVolumeMountOptions // ReadOnly indicates that the volume will be used in a readonly fashion. This // only works when the Controller has the PublishReadonly capability. ReadOnly bool // Secrets required by plugin to complete the controller publish // volume request. This field is OPTIONAL. Secrets structs.CSISecrets CSIControllerQuery }
func (*ClientCSIControllerAttachVolumeRequest) ToCSIRequest ¶ added in v0.11.0
func (c *ClientCSIControllerAttachVolumeRequest) ToCSIRequest() (*csi.ControllerPublishVolumeRequest, error)
type ClientCSIControllerAttachVolumeResponse ¶ added in v0.11.0
type ClientCSIControllerAttachVolumeResponse struct { // Opaque static publish properties of the volume. SP MAY use this // field to ensure subsequent `NodeStageVolume` or `NodePublishVolume` // calls calls have contextual information. // The contents of this field SHALL be opaque to nomad. // The contents of this field SHALL NOT be mutable. // The contents of this field SHALL be safe for the nomad to cache. // The contents of this field SHOULD NOT contain sensitive // information. // The contents of this field SHOULD NOT be used for uniquely // identifying a volume. The `volume_id` alone SHOULD be sufficient to // identify the volume. // This field is OPTIONAL and when present MUST be passed to // subsequent `NodeStageVolume` or `NodePublishVolume` calls PublishContext map[string]string }
ClientCSIControllerDetachVolumeRequest is the RPC made from the server to a Nomad client to tell a CSI controller plugin on that client to perform ControllerUnpublish for a volume on a specific client.
type ClientCSIControllerDetachVolumeRequest ¶ added in v0.11.0
type ClientCSIControllerDetachVolumeRequest struct { // The external ID of the volume to be unpublished for the node // This field is REQUIRED. VolumeID string // The CSI Node ID for the Node that the volume should be detached from. // This field is REQUIRED. This must match the NodeID that is fingerprinted // by the target node for this plugin name. ClientCSINodeID string // Secrets required by plugin to complete the controller unpublish // volume request. This field is OPTIONAL. Secrets structs.CSISecrets CSIControllerQuery }
func (*ClientCSIControllerDetachVolumeRequest) ToCSIRequest ¶ added in v0.11.0
func (c *ClientCSIControllerDetachVolumeRequest) ToCSIRequest() *csi.ControllerUnpublishVolumeRequest
type ClientCSIControllerDetachVolumeResponse ¶ added in v0.11.0
type ClientCSIControllerDetachVolumeResponse struct{}
type ClientCSIControllerValidateVolumeRequest ¶ added in v0.11.0
type ClientCSIControllerValidateVolumeRequest struct { VolumeID string // note: this is the external ID AttachmentMode structs.CSIVolumeAttachmentMode AccessMode structs.CSIVolumeAccessMode Secrets structs.CSISecrets CSIControllerQuery }
type ClientCSIControllerValidateVolumeResponse ¶ added in v0.11.0
type ClientCSIControllerValidateVolumeResponse struct { }
type ClientCSINodeDetachVolumeRequest ¶ added in v0.11.0
type ClientCSINodeDetachVolumeRequest struct { PluginID string // ID of the plugin that manages the volume (required) VolumeID string // ID of the volume to be unpublished (required) AllocID string // ID of the allocation we're unpublishing for (required) NodeID string // ID of the Nomad client targeted ExternalID string // External ID of the volume to be unpublished (required) // These fields should match the original volume request so that // we can find the mount points on the client AttachmentMode structs.CSIVolumeAttachmentMode AccessMode structs.CSIVolumeAccessMode ReadOnly bool }
ClientCSINodeDetachVolumeRequest is the RPC made from the server to a Nomad client to tell a CSI node plugin on that client to perform NodeUnpublish and NodeUnstage.
type ClientCSINodeDetachVolumeResponse ¶ added in v0.11.0
type ClientCSINodeDetachVolumeResponse struct{}
type ClientStatsResponse ¶ added in v0.8.0
ClientStatsResponse is used to return statistics about a node.
type CpuStats ¶
type CpuStats struct { SystemMode float64 UserMode float64 TotalTicks float64 ThrottledPeriods uint64 ThrottledTime uint64 Percent float64 // A list of fields whose values were actually sampled Measured []string }
CpuStats holds cpu usage related stats
type FsListRequest ¶ added in v0.8.0
type FsListRequest struct { // AllocID is the allocation to list from AllocID string // Path is the path to list Path string structs.QueryOptions }
FsListRequest is used to list an allocation's directory.
type FsListResponse ¶ added in v0.8.0
type FsListResponse struct { // Files are the result of listing a directory. Files []*AllocFileInfo structs.QueryMeta }
FsListResponse is used to return the listings of an allocation's directory.
type FsLogsRequest ¶ added in v0.8.0
type FsLogsRequest struct { // AllocID is the allocation to stream logs from AllocID string // Task is the task to stream logs from Task string // LogType indicates whether "stderr" or "stdout" should be streamed LogType string // Offset is the offset to start streaming data at. Offset int64 // Origin can either be "start" or "end" and determines where the offset is // applied. Origin string // PlainText disables base64 encoding. PlainText bool // Follow follows logs. Follow bool structs.QueryOptions }
FsLogsRequest is the initial request for accessing allocation logs.
type FsStatRequest ¶ added in v0.8.0
type FsStatRequest struct { // AllocID is the allocation to stat the file in AllocID string // Path is the path to list Path string structs.QueryOptions }
FsStatRequest is used to stat a file
type FsStatResponse ¶ added in v0.8.0
type FsStatResponse struct { // Info is the result of stating a file Info *AllocFileInfo structs.QueryMeta }
FsStatResponse is used to return the stat results of a file
type FsStreamRequest ¶ added in v0.8.0
type FsStreamRequest struct { // AllocID is the allocation to stream logs from AllocID string // Path is the path to the file to stream Path string // Offset is the offset to start streaming data at. Offset int64 // Origin can either be "start" or "end" and determines where the offset is // applied. Origin string // PlainText disables base64 encoding. PlainText bool // Limit is the number of bytes to read Limit int64 // Follow follows the file. Follow bool structs.QueryOptions }
FsStreamRequest is the initial request for streaming the content of a file.
type HealthCheckIntervalRequest ¶ added in v0.8.0
type HealthCheckIntervalRequest struct{}
type HealthCheckIntervalResponse ¶ added in v0.8.0
type HealthCheckRequest ¶ added in v0.8.0
type HealthCheckRequest struct{}
HealthCheckRequest is the request type for a type that fulfils the Health Check interface
type HealthCheckResponse ¶ added in v0.8.0
type HealthCheckResponse struct { // Drivers is a map of driver names to current driver information Drivers map[string]*structs.DriverInfo }
HealthCheckResponse is the response type for a type that fulfills the Health Check interface
func (*HealthCheckResponse) AddDriverInfo ¶ added in v0.8.0
func (h *HealthCheckResponse) AddDriverInfo(name string, driverInfo *structs.DriverInfo)
AddDriverInfo adds information about a driver to the fingerprint response. If the Drivers field has not yet been initialized, it does so here.
type MemoryStats ¶
type MemoryStats struct { RSS uint64 Cache uint64 Swap uint64 Usage uint64 MaxUsage uint64 KernelUsage uint64 KernelMaxUsage uint64 // A list of fields whose values were actually sampled Measured []string }
MemoryStats holds memory usage related stats
func (*MemoryStats) Add ¶
func (ms *MemoryStats) Add(other *MemoryStats)
type MonitorRequest ¶ added in v0.10.2
type MonitorRequest struct { // LogLevel is the log level filter we want to stream logs on LogLevel string // LogJSON specifies if log format should be unstructured or json LogJSON bool // NodeID is the node we want to track the logs of NodeID string // ServerID is the server we want to track the logs of ServerID string // PlainText disables base64 encoding. PlainText bool structs.QueryOptions }
MonitorRequest is used to request and stream logs from a client node.
type ResourceUsage ¶
type ResourceUsage struct { MemoryStats *MemoryStats CpuStats *CpuStats DeviceStats []*device.DeviceGroupStats }
ResourceUsage holds information related to cpu and memory stats
func (*ResourceUsage) Add ¶
func (ru *ResourceUsage) Add(other *ResourceUsage)
type RpcError ¶ added in v0.8.0
RpcError is used for serializing errors with a potential error code
func NewRpcError ¶ added in v0.8.0
type StreamErrWrapper ¶ added in v0.8.0
type StreamErrWrapper struct { // Error stores any error that may have occurred. Error *RpcError // Payload is the payload Payload []byte }
StreamErrWrapper is used to serialize output of a stream of a file or logs.
type TaskResourceUsage ¶
type TaskResourceUsage struct { ResourceUsage *ResourceUsage Timestamp int64 // UnixNano Pids map[string]*ResourceUsage }
TaskResourceUsage holds aggregated resource usage of all processes in a Task and the resource usage of the individual pids