TracingService Plugin
Applications that consist of multiple services can be difficult to debug, as a single request can span multiple services. Distributed tracing tells the story of your request as it is processed through your system. Distributed tracing is a powerful tool to debug and analyze your system in addition to request logging and metrics.
The TracingService
When enabled, the TracingService
will instruct Ambassador Edge Stack to initiate a trace on requests by generating and populating an x-request-id
HTTP header. Services can make use of this x-request-id
header in logging and forward it in downstream requests for tracing. Ambassador Edge Stack also integrates with external trace visualization services, including LightStep and Zipkin-compatible APIs such as Zipkin and Jaeger to allow you to store and visualize traces. You can read further on Envoy's Tracing capabilities.
A TracingService
manifest configures Ambassador Edge Stack to use an external trace visualization service:
---apiVersion: getambassador.io/v2kind: TracingServicemetadata:name: tracingspec:service: "example-zipkin:9411"driver: zipkinconfig: {}tag_headers:- ":authority"- ":path"sampling:overall: 100
service
gives the URL of the external HTTP trace service.driver
provides the driver information that handles communicating with theservice
. Supported values arelightstep
,zipkin
, anddatadog
.config
provides additional configuration options for the selecteddriver
.tag_headers
(optional) if present, specifies a list of other HTTP request headers which will be used as tags in the trace's span.sampling
(optional) if present, specifies some target percentages of requests that will be traced.client
: percentage of requests that will be force traced if thex-client-trace-id
header is set. Defaults to 100.random
: percentage of requests that will be randomly traced. Defaults to 100.overall
: percentage of requests that will be traced after all other checks have been applied (force tracing, sampling, etc.). This field functions as an upper limit on the total configured sampling rate. For instance, settingclient
to100%
butoverall
to1%
will result in only1%
of client requests with the appropriate headers to be force traced. Defaults to 100.
Please note that you must use the HTTP/2 pseudo-header names. For example:
- the
host
header should be specified as the:authority
header; and - the
method
header should be specified as the:method
header.
lightstep
Driver Configurations
access_token_file
provides the location of the file containing the access token to the LightStep API.
zipkin
Driver Configurations
collector_endpoint
gives the API endpoint of the Zipkin service where the spans will be sent. The default value is/api/v1/spans
collector_endpoint_version
gives the API version Envoy will use when sending data to your Zipkin collector. The default value isHTTP_JSON_V1
collector_endpoint_hostname
sets the hostname Envoy will use when sending data to your Zipkin collector. The default value is the name of the underlying Envoy cluster.trace_id_128bit
whether a 128-bittrace id
will be used when creating a new trace instance. Defaults totrue
. Setting tofalse
will result in a 64-bit trace id being used.shared_span_context
whether client and server spans will share the samespan id
. The default value istrue
.
datadog
Driver Configurations
service_name
the name of the service which is attached to the traces. The default value isambassador
.
You may only use a single TracingService
manifest per Ambassador deployment. Ensure ambassador_id is set correctly in the TracingService
manifest.
Example
Check out the DataDog and Zipkin HOWTOs.
Questions?
We’re here to help. If you have questions, join our Slack or contact us.