pub struct LogRecord {
pub seq_no: u64,
pub epoch: u64,
pub timestamp: Timestamp,
pub local_worker: Worker,
pub activity_type: ActivityType,
pub event_type: EventType,
pub remote_worker: Option<Worker>,
pub operator_id: Option<OperatorId>,
pub channel_id: Option<ChannelId>,
pub correlator_id: Option<u64>,
pub length: Option<usize>,
}
Expand description
A LogRecord
constitutes the unified struct
representation of
log messages from various stream processors.
It is the underlying structure from which the PAG construction starts.
If necessary, it can also be serialized e.g. into a msgpack
representation.
Fields§
§seq_no: u64
worker-unique identifier of a message, given in order the events are logged in the computation.
epoch: u64
epoch of the computation this record belongs to
timestamp: Timestamp
Event time in nanoseconds since the Epoch (midnight, January 1, 1970 UTC).
local_worker: Worker
Context this event occured in; denotes which of the parallel timelines it belongs to.
activity_type: ActivityType
Describes the instrumentation point which triggered this event.
event_type: EventType
Identifies which end of an edge this program event belongs to.
remote_worker: Option<Worker>
Similar to local_worker
but specifies the worker ID for the other end of a sent/received message.
operator_id: Option<OperatorId>
Unique id for the operator in the dataflow. This only applies for some event types, e.g. scheduling or processing.
channel_id: Option<ChannelId>
Unique id for the channel in the dataflow. This only applies for some event types, e.g. data / control messages.
correlator_id: Option<u64>
correlates remote events belonging together
length: Option<usize>
Number of records to detect skew