Batch Processor
🚧 This document is work in progress.
This document uses key words such as "MUST", "SHOULD", and "MAY" as defined in RFC 2119 to indicate requirement levels.
The BatchProcessor batches spans and logs into one envelope to reduce the number of HTTP requests. When an SDK implements span streaming or logs, it MUST use a BatchProcessor, which is similar to OpenTelemetry's Batch Processor. The BatchProcessor holds logs and finished spans in memory and batches them together into envelopes. It uses a combination of time and size-based batching. When writing this, the BatchProcessor only handles spans and logs, but an SDK MAY use it for other telemetry data in the future.
Whenever the SDK finishes a span or captures a log, it MUST put it into the BatchProcessor. The SDK MUST NOT put unfinished spans into the BatchProcessor.
The BatchProcessor MUST start a timeout of 5 seconds when the SDK adds the first span or log. When the timeout exceeds, the BatchProcessor MUST send all spans or logs, no matter how many items it contains. The SDK MAY choose a different value for the timeout, but it MUST NOT exceed 30 seconds, as this can lead to problems with the span buffer on the backend, which uses a time interval of 60 seconds for determining segments for spans.
The BatchProcessor MUST send all items after the SDK when containing spans or logs exceeding 1MiB in size. The SDK MAY choose a different value for the max batch size keeping the envelope max sizes in mind. The SDK MUST calculate the size of a span or a log to manage the BatchProcessor's memory footprint. The SDK MUST serialize the span or log and calculate the size based on the serialized JSON bytes. As serialization is expensive, the BatchProcessor SHOULD keep track of the serialized spans and logs and pass these to the envelope to avoid serializing multiple times.
When the BatchProcessor sends all spans or logs, it MUST reset its timeout and remove all spans and logs. The SDK MUST apply filtering and sampling before adding spans or logs to the BatchProcessor. The SDK MUST apply rate limits to spans and logs after they leave the BatchProcessor to send as much data as possible by dropping data as late as possible.
The detailed specification is written in the Gherkin syntax. The specification uses spans as an example, but the same applies to logs or any other future telemetry data.
Scenario: No spans in BatchProcessor 1 span added
Given no spans in the BatchProcessor
When the SDK finishes 1 span
Then the SDK puts this span to the BatchProcessor
And starts a timeout of 5 seconds
And doesn't send the span to Sentry
Scenario: Span added before timeout exceeds
Given span A in the BatchProcessor
Given 4.9 seconds pass
When the SDK finishes span B
Then the SDK adds span B to the BatchProcessor
And doesn't reset the timeout
And doesn't send the spans A and B in the BatchProcessor to Sentry
Scenario: Spans with size of 1 MiB - 1 byte added, timeout exceeds
Given spans with size of 1 MiB - 1 byte in the BatchProcessor
When the timeout exceeds
Then the SDK adds all the spans to one envelope
And sends them to Sentry
And resets the timeout
And clears the BatchProcessor
Scenario: Spans with size of 1 MiB - 1 byte added within 4.9 seconds
Given spans with size of 1 MiB - 1 byte in the BatchProcessor
When the SDK finishes another span and puts it into the BatchProcessor
Then the BatchProcessor puts all spans into one envelope
And sends the envelope to Sentry
And resets the timeout
And clears the BatchProcessor
Scenario: Unfinished spans
Given no span is in the BatchProcessor
When the SDK starts a span but doesn't finish it
Then the BatchProcessor is empty
Scenario: Span filtered out
Given no span is in the BatchProcessor
When the finishes a span
And the span is filtered out
Then the BatchProcessor is empty
Scenario: Span not sampled
Given no span is in the BatchProcessor
When the finishes a span
And the span is not sampled
Then the BatchProcessor is empty
Scenario: 1 span added application crashes
Given 1 span in the SpansAggregator
When the SDK detects a crash
Then the SDK does nothing with the items in the BatchProcessor
And loses the spans in the BatchProcessor
Our documentation is open source and available on GitHub. Your contributions are welcome, whether fixing a typo (drat!) or suggesting an update ("yeah, this would be better").