Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type BasicOperator ¶
type BasicOperator struct {
// contains filtered or unexported fields
}
func NewBasicOperator ¶
func NewBasicOperator(conn config.ConnectionGetter, env map[string]string) (*BasicOperator, error)
func (*BasicOperator) Run ¶
func (op *BasicOperator) Run(ctx context.Context, ti scheduler.TaskInstance) error
type Config ¶
type Config struct {
ServiceAccountJSON string `yaml:"service_account_json,omitempty"`
ServiceAccountFile string `yaml:"service_account_file,omitempty"`
UseApplicationDefaultCredentials bool `yaml:"use_application_default_credentials,omitempty"`
ProjectID string `yaml:"project_id,omitempty"`
Region string `yaml:"region" json:"region"`
Workspace string `yaml:"workspace"`
ExecutionRole string `yaml:"execution_role"`
SubnetworkURI string `yaml:"subnetwork_uri,omitempty"`
NetworkTags []string `yaml:"network_tags,omitempty"`
KmsKey string `yaml:"kms_key,omitempty"`
StagingBucket string `yaml:"staging_bucket,omitempty"`
MetastoreService string `yaml:"metastore_service,omitempty"`
}
type GCSLogConsumer ¶ added in v0.11.649
type GCSLogConsumer struct {
// contains filtered or unexported fields
}
GCSLogConsumer streams the Spark driver output that Dataproc Serverless writes to the batch's staging bucket (RuntimeInfo.OutputUri). This is the same source that `gcloud dataproc batches wait` streams.
We deliberately do not read job logs from Cloud Logging: Dataproc Serverless writes the driver/executor stdout, stderr and log4j output to this GCS location, and that output does not reliably appear in Cloud Logging (only control-plane logs such as the autoscaler do).
func (*GCSLogConsumer) Flush ¶ added in v0.11.649
func (l *GCSLogConsumer) Flush() []LogLine
Flush returns any remaining buffered output, including a trailing line that is not newline-terminated. It should be called once after the batch reaches a terminal state.
func (*GCSLogConsumer) Next ¶ added in v0.11.649
func (l *GCSLogConsumer) Next() []LogLine
Next returns any complete log lines written since the last call.
func (*GCSLogConsumer) SetOutputURI ¶ added in v0.11.649
func (l *GCSLogConsumer) SetOutputURI(uri string)
SetOutputURI records the GCS location of the driver output. Dataproc only populates RuntimeInfo.OutputUri once the batch starts running, so this is called on every poll until a non-empty URI is observed.
type JobRunParams ¶
type JobRunParams struct {
Project string
Region string
RuntimeVersion string
ContainerImage string
Config string
Args []string
Timeout time.Duration
Workspace string
ExecutionRole string
SubnetworkURI string
NetworkTags []string
KmsKey string
StagingBucket string
MetastoreService string
Cohort string
AutotuningScenarios []dataprocpb.AutotuningConfig_Scenario
}
type LogConsumer ¶
type LogConsumer interface {
// Next returns log lines produced since the previous call.
Next() []LogLine
// Flush returns any remaining buffered output, including a final line not
// terminated by a newline.
Flush() []LogLine
// SetOutputURI points the consumer at the GCS driver-output location.
SetOutputURI(uri string)
}
type MissingFieldsError ¶
type MissingFieldsError struct {
Fields []string
}
func (*MissingFieldsError) Error ¶
func (e *MissingFieldsError) Error() string
type NoOpLogConsumer ¶
type NoOpLogConsumer struct{}
NoOpLogConsumer is a log consumer that does nothing.
func (*NoOpLogConsumer) Flush ¶ added in v0.11.649
func (l *NoOpLogConsumer) Flush() []LogLine
func (*NoOpLogConsumer) Next ¶
func (l *NoOpLogConsumer) Next() []LogLine
func (*NoOpLogConsumer) SetOutputURI ¶ added in v0.11.649
func (l *NoOpLogConsumer) SetOutputURI(string)