Pinpoint

Version-1.6.0 · Experimental

The built-in Pinpoint Agent in Datakit is used to receive, calculate, and analyze Pinpoint Tracing protocol data.

Configuration¶

Collector Configuration¶

[[inputs.pinpoint]]
  ## Pinpoint service endpoint for
  ## - Span Server
  ## - Agent Server(unimplemented, for service intactness and compatibility)
  ## - Metadata Server(unimplemented, for service intactness and compatibility)
  ## - Profiler Server(unimplemented, for service intactness and compatibility)
  address = "127.0.0.1:9991"

  ## Keep rare tracing resources list switch.
  ## If some resources are rare enough(not presend in 1 hour), those resource will always send
  ## to data center and do not consider samplers and filters.
  # keep_rare_resource = false

  ## delete trace message
  # del_message = true

  ## Ignore tracing resources map like service:[resources...].
  ## The service name is the full service name in current application.
  ## The resource list is regular expressions uses to block resource names.
  ## If you want to block some resources universally under all services, you can set the
  ## service name as "*". Note: double quotes "" cannot be omitted.
  # [inputs.pinpoint.close_resource]
    # service1 = ["resource1", "resource2", ...]
    # service2 = ["resource1", "resource2", ...]
    # "*" = ["close_resource_under_all_services"]
    # ...

  ## Sampler config uses to set global sampling strategy.
  ## sampling_rate used to set global sampling rate.
  # [inputs.pinpoint.sampler]
    # sampling_rate = 1.0

  # [inputs.pinpoint.tags]
    # key1 = "value1"
    # key2 = "value2"
    # ...

  ## Storage config a local storage space in hard dirver to cache trace data.
  ## path is the local file path used to cache data.
  ## capacity is total space size(MB) used to store data.
  # [inputs.pinpoint.storage]
    # path = "./pinpoint_storage"
    # capacity = 5120

# Pinpoint GRPC service endpoint for
# - Span Server
# - Agent Server(unimplemented, for service intactness and compatibility)
# - Metadata Server(unimplemented, for service intactness and compatibility)
# - Profiler Server(unimplemented, for service intactness and compatibility)
address = "127.0.0.1:9991"

Pinpoint Agent configuration¶

• Download the required Pinpoint APM Agent

Pinpoint supports the multi-language APM Collector. This document uses JAVA Agent for configuration. Download JAVA APM Collector.

• Configure Pinpoint APM Collector, open /path_to_pinpoint_agent/pinpoint-root.config and configure the corresponding multi-service ports

  • Configure profiler.transport.module = GRPC
  • Configure profiler.transport.grpc.agent.collector.port = 9991 (i.e. the port configured in Datakit Pinpoint Agent)
  • Configure profiler.transport.grpc.metadata.collector.port = 9991(i.e. the port configured in Datakit Pinpoint Agent)
  • Configure profiler.transport.grpc.stat.collector.port = 9991 (i.e. the port configured in Datakit Pinpoint Agent)
  • Configure profiler.transport.grpc.span.collector.port = 9991 (i.e. the port configured in Datakit Pinpoint Agent)

• Start Pinpoint APM Agent startup command

$ java -javaagent:/path_to_pinpoint/pinpoint-bootstrap.jar \
    -Dpinpoint.agentId=agent-id \
    -Dpinpoint.applicationName=app-name \
    -Dpinpoint.config=/path_to_pinpoint/pinpoint-root.config \
    -jar /path_to_your_app.jar

Datakit link data follows the OpenTracing protocol. A link in Datakit is concatenated through a simple parent-child (the child span stores the id of the parent span) structure and each span corresponds to a function call.

Pinpoint APM link data is more complex:

• The parent span is responsible for generating the ID of the child span
• The ID of the parent span must also be stored in the child span.
• Use span event instead of span in OpenTracing
• A span is a response process for a service

PinPointV2¶

DataKit 1.19.0 version has been re-optimized and changed source to PinPointV2. The new version of link data reorganizes the relationship between SpanChunk and Span, the relationship between Event and Span, and the relationship between Span and Span. And the time alignment problem between startElapsed and endElapsed in Event.

Main logical points:

• Cache the serviceType service table and write it to a file to prevent data loss when DataKit restarts.
• Cache if parentSpanId in Span is not -1. For example, if parentSpanId:-1 is used, the Span will be fetched from the cache and spliced into a link based on the nextSpanId in spanEvent.
• Cache all event in SpanChunk, until the main Span is received, all are taken out from the cache and appended to the link.
• Accumulate startElapsed in the current Event in order as the start time of the next Event.
• Determine the parent-child relationship of the current Event according to the Depth field.
• Database queries will replace the current 'resource' name with sql statements.

Tracing¶

pinpoint¶

• Tags

• Metrics

Metric¶

pinpoint-metric¶

• Tags

• Metrics

Pinpoint References¶

• Pinpoint official documentation
• Pinpoint version documentation library
• Pinpoint official repository
• Pinpoint online example