-
Notifications
You must be signed in to change notification settings - Fork 64
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Details attributes and http header encoding #8
Changes from 1 commit
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1,2 +1,55 @@ | ||
# b3-java | ||
Shared library for B3 Propagation in text and binary representation. | ||
# b3-propagation | ||
|
||
This repository elaborates identifiers used to place an operation in a trace tree. These attributes are propagated in-process, and eventually downstream (often via http headers), to ensure all activity originating from the same root are collected together. A sampling decision is made at the root of the trace, and this indicates if trace details should be collected and reported to the tracing system (usually Zipkin) or not. | ||
|
||
# Identifiers | ||
Trace identifiers are 64 or 128-bit, but all span identifiers within a trace are 64-bit. All identifiers are opaque. | ||
|
||
## TraceId | ||
|
||
The TraceId is 64 or 128-bit in length and indicates the overall ID of the trace. Every span in a trace shares this ID. | ||
|
||
### Http Header format | ||
The `X-B3-TraceId` header is required and is encoded as 32 or 16 hex characters. For example, a 128-bit TraceId header might look like: `X-B3-TraceId: 463ac35c9f6413ad48485a3953bb6124` | ||
|
||
## SpanId | ||
|
||
The SpanId is 64-bit in length and indicates the position of the current operation in the trace tree. The value should not be interpreted: it may or may not be derived from the value of the TraceId. | ||
|
||
### Http Header format | ||
The `X-B3-SpanId` header is required and is encoded as 16 hex characters. For example, a SpanId header might look like: `X-B3-SpanId: a2fb4a1d1a96d312` | ||
|
||
## ParentSpanId | ||
|
||
The ParentSpanId is 64-bit in length and indicates the position of the parent operation in the trace tree. When the span is the root of the trace tree, the ParentSpanId is absent. | ||
|
||
### Http Header format | ||
The `X-B3-ParentSpanId` header must be present on a child span and absent on the root span. It is encoded as 16 hex characters. For example, a ParentSpanId header might look like: `X-B3-ParentSpanId: 0020000000000001` | ||
|
||
# Flags | ||
The following flags are reported either in a flag set or separate attributes. | ||
|
||
## Sampled | ||
|
||
When the Sampled is 1, report this span to the tracing system. When it is 0, do do not. When B3 attributes are sent without the Sampled attribute, the receiver should make the decision. Once Sampled is set to 0 or 1, the value should be consistently sent downstream. | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. "When the Sampled is 1" sounds strange to me. Maybe "When Sampled is 1" or "When the Sampled attribute is 1" There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Thanks.. I think I meant |
||
|
||
### Http Header format | ||
The `X-B3-Sampled` header is encoded as "1" or "0". Absent means defer the decision to the receiver of this header. For example, a Sampled header might look like: `X-B3-Sampled: 1` | ||
|
||
### Details | ||
|
||
It may not be obvious why you'd send Sampled=0 to the next hop. Imagine a service decides not to trace an operation and makes 2 out-going calls, and these branched out further. If 0 ("don't trace") isn't propagated, the system might receive only parts of the operation, confusing users. | ||
|
||
Leaving Sampled absent is special-case. The only known use-cases are the following: | ||
|
||
* Debug trace: When setting Flags to 1, sampling is implicit | ||
* Externally provisioned IDs: When you want to control IDs, but not sampling policy | ||
|
||
Unless it is a debug trace, leaving sampled unset is typically for ID correlation. For example, someone re-uses a global identifier from another system, or correlating in logs. In these cases, the caller knows the ID they want, but allows the next hop to decide if it will be traced or not. The caller should not report a span to the tracing system using this ID unless they propagate Sampled=1. | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. cc @jcarres-mdsol @dangets This particularly relates to functionality in Finagle, which is missing in Brave openzipkin/brave#277 |
||
|
||
## Debug | ||
When Debug is set, the trace should be reported to the tracing system and also override any collection-tier sampling policy. Debug implies Sampled. | ||
|
||
### Http Header format | ||
Debug is encoded as `X-B3-Flags: 1`. Since Debug implies Sampled, so don't also send "X-B3-Sampled: 1". | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
"When it is 0, do do not." → "When it is 0, do not."