warehouse

package
v1.0.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Oct 14, 2021 License: AGPL-3.0 Imports: 58 Imported by: 0

Documentation

Index

Constants

View Source 
const (
	Waiting                   = "waiting"
	GeneratedUploadSchema     = "generated_upload_schema"
	CreatedTableUploads       = "created_table_uploads"
	GeneratedLoadFiles        = "generated_load_files"
	UpdatedTableUploadsCounts = "updated_table_uploads_counts"
	CreatedRemoteSchema       = "created_remote_schema"
	ExportedUserTables        = "exported_user_tables"
	ExportedData              = "exported_data"
	ExportedIdentities        = "exported_identities"
	Aborted                   = "aborted"
)

Upload Status

View Source 
const (
	GeneratingStagingFileFailedState        = "generating_staging_file_failed"
	GeneratedStagingFileState               = "generated_staging_file"
	PopulatingHistoricIdentitiesState       = "populating_historic_identities"
	PopulatingHistoricIdentitiesStateFailed = "populating_historic_identities_failed"
	FetchingRemoteSchemaFailed              = "fetching_remote_schema_failed"
	InternalProcessingFailed                = "internal_processing_failed"
)
View Source 
const (
	TableUploadExecuting               = "executing"
	TableUploadUpdatingSchema          = "updating_schema"
	TableUploadUpdatingSchemaFailed    = "updating_schema_failed"
	TableUploadUpdatedSchema           = "updated_schema"
	TableUploadExporting               = "exporting_data"
	TableUploadExportingFailed         = "exporting_data_failed"
	UserTableUploadExportingFailed     = "exporting_user_tables_failed"
	IdentityTableUploadExportingFailed = "exporting_identities_failed"
	TableUploadExported                = "exported_data"
)

Table Upload status

View Source 
const (
	UploadStatusField          = "status"
	UploadStartLoadFileIDField = "start_load_file_id"
	UploadEndLoadFileIDField   = "end_load_file_id"
	UploadUpdatedAtField       = "updated_at"
	UploadTimingsField         = "timings"
	UploadSchemaField          = "schema"
	MergedSchemaField          = "mergedschema"
	UploadLastExecAtField      = "last_exec_at"
)
View Source 
const (
	MasterMode        = "master"
	SlaveMode         = "slave"
	MasterSlaveMode   = "master_and_slave"
	EmbeddedMode      = "embedded"
	PooledWHSlaveMode = "embedded_master"
)

warehouses worker modes

View Source 
const (
	DegradedMode                  = "degraded"
	StagingFilesPGNotifierChannel = "process_staging_file"
)
View Source 
const (
	CloudSourceCateogry = "cloud"
)

Variables

View Source 
var (
	WarehouseDestinations []string

	ShouldForceSetLowerVersion bool
)

Functions

func CheckCurrentTimeExistsInExcludeWindow added in v0.1.10 

func CheckCurrentTimeExistsInExcludeWindow(currentTime time.Time, windowStartTime string, windowEndTime string) bool

func CheckForWarehouseEnvVars added in v0.1.10 

func CheckForWarehouseEnvVars() bool

CheckForWarehouseEnvVars Checks if all the required Env Variables for Warehouse are present

func CheckPGHealth 

func CheckPGHealth(dbHandle *sql.DB) bool

func GetExludeWindowStartEndTimes added in v0.1.10 

func GetExludeWindowStartEndTimes(excludeWindow map[string]interface{}) (string, string)

func GetPrevScheduledTime 

func GetPrevScheduledTime(syncFrequency, syncStartAt string, currTime time.Time) time.Time

GetPrevScheduledTime returns closest previous scheduled time eg. Syncing every 3hrs starting at 13:00 (scheduled times: 13:00, 16:00, 19:00, 22:00, 01:00, 04:00, 07:00, 10:00) prev scheduled time for current time (eg. 18:00 -> 16:00 same day, 00:30 -> 22:00 prev day)

func InitWarehouseAPI added in v0.1.10 

func InitWarehouseAPI(dbHandle *sql.DB, log logger.LoggerI)

func ScheduledTimes 

func ScheduledTimes(syncFrequency, syncStartAt string) []int

ScheduledTimes returns all possible start times (minutes from start of day) as per schedule eg. Syncing every 3hrs starting at 13:00 (scheduled times: 13:00, 16:00, 19:00, 22:00, 01:00, 04:00, 07:00, 10:00)

func Start 

func Start(app app.Interface)

func TriggerUploadHandler added in v0.1.10 

func TriggerUploadHandler(sourceID string, destID string) error

Types

type BatchRouterEventT added in v0.1.10 

type BatchRouterEventT struct {
	Metadata MetadataT `json:"metadata"`
	Data     DataT     `json:"data"`
}

type ColumnInfoT added in v0.1.10 

type ColumnInfoT struct {
	ColumnVal  interface{}
	ColumnType string
}

type DataT added in v0.1.10 

type DataT map[string]interface{}

type ErrorResponseT 

type ErrorResponseT struct {
	Error string
}

type HandleT 

type HandleT struct {
	// contains filtered or unexported fields
}

func (*HandleT) Disable 

func (wh *HandleT) Disable()

Disable disables a router:)

func (*HandleT) Enable 

func (wh *HandleT) Enable()

Enable enables a router :)

func (*HandleT) Setup 

func (wh *HandleT) Setup(whType string, whName string)

type JobRunT added in v0.1.10 

type JobRunT struct {
	// contains filtered or unexported fields
}

Temporary store for processing staging file to load file

func (*JobRunT) GetWriter added in v0.1.10 

func (jobRun *JobRunT) GetWriter(tableName string) (warehouseutils.LoadFileWriterI, error)

type LoadFileJobT 

type LoadFileJobT struct {
	Upload                     UploadT
	StagingFile                *StagingFileT
	Schema                     map[string]map[string]string
	Warehouse                  warehouseutils.WarehouseT
	Wg                         *misc.WaitGroup
	LoadFileIDsChan            chan []int64
	TableToBucketFolderMap     map[string]string
	TableToBucketFolderMapLock *sync.RWMutex
}

type MetadataT added in v0.1.10 

type MetadataT struct {
	Table        string            `json:"table"`
	Columns      map[string]string `json:"columns"`
	IsMergeRule  bool              `json:"isMergeRule"`
	ReceivedAt   time.Time         `json:"receivedAt"`
	MergePropOne string            `json:"mergePropOne"`
	MergePropTwo string            `json:"mergePropTwo"`
}

type PayloadT 

type PayloadT struct {
	BatchID              string
	UploadID             int64
	StagingFileID        int64
	StagingFileLocation  string
	UploadSchema         map[string]map[string]string
	SourceID             string
	SourceName           string
	DestinationID        string
	DestinationName      string
	DestinationType      string
	DestinationNamespace string
	DestinationConfig    interface{}
	UniqueLoadGenID      string
	UseRudderStorage     bool
	RudderStoragePrefix  string
	Output               []loadFileUploadOutputT
	LoadFilePrefix       string // prefix for the load file name
	LoadFileType         string
}

type ProcessStagingFilesJobT 

type ProcessStagingFilesJobT struct {
	Upload    UploadT
	List      []*StagingFileT
	Warehouse warehouseutils.WarehouseT
}

type QueryInput added in v0.1.10 

type QueryInput struct {
	DestID       string
	SourceID     string
	SQLStatement string
}

type SchemaHandleT added in v0.1.10 

type SchemaHandleT struct {
	// contains filtered or unexported fields
}

type StagingFileT 

type StagingFileT struct {
	ID               int64
	Location         string
	SourceID         string
	Schema           json.RawMessage
	Status           string // enum
	CreatedAt        time.Time
	FirstEventAt     time.Time
	LastEventAt      time.Time
	UseRudderStorage bool
	// cloud sources specific info
	SourceBatchID   string
	SourceTaskID    string
	SourceTaskRunID string
	SourceJobID     string
	SourceJobRunID  string
	TimeWindow      time.Time
}

type TableSkipError added in v0.1.10 

type TableSkipError struct {
	// contains filtered or unexported fields
}

TableSkipError is a custom error type to capture if a table load is skipped because of a previously failed table load

func (*TableSkipError) Error added in v0.1.10 

func (tse *TableSkipError) Error() string

type TableUploadReqT added in v0.1.10 

type TableUploadReqT struct {
	UploadID int64
	Name     string
	API      UploadAPIT
}

func (TableUploadReqT) GetWhTableUploads added in v0.1.10 

func (tableUploadReq TableUploadReqT) GetWhTableUploads() ([]*proto.WHTable, error)

type TableUploadResT added in v0.1.10 

type TableUploadResT struct {
	ID         int64     `json:"id"`
	UploadID   int64     `json:"upload_id"`
	Name       string    `json:"name"`
	Error      string    `json:"error"`
	Status     string    `json:"status"`
	Count      int32     `json:"count"`
	LastExecAt time.Time `json:"last_exec_at"`
	Duration   int32     `json:"duration"`
}

type TableUploadStatusT added in v0.1.10 

type TableUploadStatusT struct {
	// contains filtered or unexported fields
}

TableUploadStatusT captures the status of each table upload along with its parent upload_job's info like destionation_id and namespace

type TableUploadT added in v0.1.10 

type TableUploadT struct {
	// contains filtered or unexported fields
}

func NewTableUpload added in v0.1.10 

func NewTableUpload(uploadID int64, tableName string) *TableUploadT

type TablesResT added in v0.1.10 

type TablesResT struct {
	Tables []TableUploadResT `json:"tables,omitempty"`
}

type UploadAPIT added in v0.1.10 

type UploadAPIT struct {
	// contains filtered or unexported fields
}
var UploadAPI UploadAPIT

type UploadColumnT added in v0.1.10 

type UploadColumnT struct {
	Column string
	Value  interface{}
}

type UploadColumnsOpts added in v0.1.10 

type UploadColumnsOpts struct {
	Fields []UploadColumnT
	Txn    *sql.Tx
}

type UploadJobT added in v0.1.10 

type UploadJobT struct {
	// contains filtered or unexported fields
}

func (*UploadJobT) GetLoadFileGenStartTIme added in v0.1.10 

func (job *UploadJobT) GetLoadFileGenStartTIme() time.Time

func (*UploadJobT) GetLoadFileType added in v0.1.10 

func (job *UploadJobT) GetLoadFileType() string

func (*UploadJobT) GetLoadFilesMetadata added in v0.1.10 

func (job *UploadJobT) GetLoadFilesMetadata(options warehouseutils.GetLoadFilesOptionsT) (loadFiles []warehouseutils.LoadFileT)

func (*UploadJobT) GetLocalSchema added in v0.1.10 

func (job *UploadJobT) GetLocalSchema() warehouseutils.SchemaT

func (*UploadJobT) GetSampleLoadFileLocation added in v0.1.10 

func (job *UploadJobT) GetSampleLoadFileLocation(tableName string) (location string, err error)

func (*UploadJobT) GetSchemaInWarehouse added in v0.1.10 

func (job *UploadJobT) GetSchemaInWarehouse() (schema warehouseutils.SchemaT)

func (*UploadJobT) GetSingleLoadFile added in v0.1.10 

func (job *UploadJobT) GetSingleLoadFile(tableName string) (warehouseutils.LoadFileT, error)

func (*UploadJobT) GetTableSchemaInUpload added in v0.1.10 

func (job *UploadJobT) GetTableSchemaInUpload(tableName string) warehouseutils.TableSchemaT

func (*UploadJobT) GetTableSchemaInWarehouse added in v0.1.10 

func (job *UploadJobT) GetTableSchemaInWarehouse(tableName string) warehouseutils.TableSchemaT

func (*UploadJobT) ShouldOnDedupUseNewRecord added in v0.1.10 

func (job *UploadJobT) ShouldOnDedupUseNewRecord() bool

func (*UploadJobT) UpdateLocalSchema added in v0.1.10 

func (job *UploadJobT) UpdateLocalSchema(schema warehouseutils.SchemaT) error

func (*UploadJobT) UseRudderStorage added in v0.1.10 

func (job *UploadJobT) UseRudderStorage() bool

type UploadPagination added in v0.1.10 

type UploadPagination struct {
	Total  int32 `json:"total"`
	Limit  int32 `json:"limit"`
	Offset int32 `json:"offset"`
}

type UploadReqT added in v0.1.10 

type UploadReqT struct {
	WorkspaceID string
	UploadId    int64
	API         UploadAPIT
}

func (UploadReqT) GetWHUpload added in v0.1.10 

func (uploadReq UploadReqT) GetWHUpload() (*proto.WHUploadResponse, error)

func (UploadReqT) TriggerWHUpload added in v0.1.10 

func (uploadReq UploadReqT) TriggerWHUpload() error

type UploadResT added in v0.1.10 

type UploadResT struct {
	ID              int64             `json:"id"`
	Namespace       string            `json:"namespace"`
	SourceID        string            `json:"source_id"`
	DestinationID   string            `json:"destination_id"`
	DestinationType string            `json:"destination_type"`
	Status          string            `json:"status"`
	Error           string            `json:"error"`
	Attempt         int32             `json:"attempt"`
	Duration        int32             `json:"duration"`
	NextRetryTime   string            `json:"nextRetryTime"`
	FirstEventAt    time.Time         `json:"first_event_at"`
	LastEventAt     time.Time         `json:"last_event_at"`
	Tables          []TableUploadResT `json:"tables,omitempty"`
}

type UploadStatusOpts added in v0.1.10 

type UploadStatusOpts struct {
	Status           string
	AdditionalFields []UploadColumnT
	ReportingMetric  types.PUReportedMetric
}

type UploadT added in v0.1.10 

type UploadT struct {
	ID                   int64
	Namespace            string
	SourceID             string
	SourceType           string
	SourceCategory       string
	DestinationID        string
	DestinationType      string
	StartStagingFileID   int64
	EndStagingFileID     int64
	StartLoadFileID      int64
	EndLoadFileID        int64
	Status               string
	UploadSchema         warehouseutils.SchemaT
	MergedSchema         warehouseutils.SchemaT
	Error                json.RawMessage
	Timings              []map[string]string
	FirstAttemptAt       time.Time
	LastAttemptAt        time.Time
	Attempts             int64
	Metadata             json.RawMessage
	FirstEventAt         time.Time
	LastEventAt          time.Time
	UseRudderStorage     bool
	LoadFileGenStartTime time.Time
	TimingsObj           sql.NullString
	Priority             int
	// cloud sources specific info
	SourceBatchID   string
	SourceTaskID    string
	SourceTaskRunID string
	SourceJobID     string
	SourceJobRunID  string
	LoadFileType    string
}

type UploadsReqT added in v0.1.10 

type UploadsReqT struct {
	WorkspaceID     string
	SourceID        string
	DestinationID   string
	DestinationType string
	Status          string
	Limit           int32
	Offset          int32
	API             UploadAPIT
}

func (*UploadsReqT) GetWhUploads added in v0.1.10 

func (uploadsReq *UploadsReqT) GetWhUploads() (uploadsRes *proto.WHUploadsResponse, err error)

func (*UploadsReqT) TriggerWhUploads added in v0.1.10 

func (uploadsReq *UploadsReqT) TriggerWhUploads() (err error)

type UploadsResT added in v0.1.10 

type UploadsResT struct {
	Uploads    []UploadResT     `json:"uploads"`
	Pagination UploadPagination `json:"pagination"`
}

type WarehouseAdmin added in v0.1.10 

type WarehouseAdmin struct{}

func (*WarehouseAdmin) Query added in v0.1.10 

func (wh *WarehouseAdmin) Query(s QueryInput, reply *warehouseutils.QueryResult) error

Query the underlying warehouse

func (*WarehouseAdmin) QueryWhTables added in v0.1.10 

func (wh *WarehouseAdmin) QueryWhTables(tableUploadReq TableUploadReqT, reply *[]byte) error

func (*WarehouseAdmin) QueryWhUploads added in v0.1.10 

func (wh *WarehouseAdmin) QueryWhUploads(uploadsReq UploadsReqT, reply *[]byte) error

func (*WarehouseAdmin) TriggerUpload added in v0.1.10 

func (wh *WarehouseAdmin) TriggerUpload(off bool, reply *string) error

TriggerUpload sets uploads to start without delay

Source Files

View all Source files

Directories

Path Synopsis
schema-repository

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.