Migrating compound extraction span link tests to weblog #3499
Conversation
mhlidd
changed the title
mhlidd
requested review from
Kyle-Verhoog,
avara1986,
smola and
jandro996
and removed request for
a team
| assert link1["trace_id"] == 1 | ||
| assert link1["span_id"] == 987654321 | ||
| assert link1["attributes"] == {"reason": "terminated_context", "context_headers": "tracecontext"} | ||
| assert link1["tracestate"] == "dd=s:2;t.tid:1111111111111111,foo=1" |
There was a problem hiding this comment.
nit: Assertion on tracestate could get tricky. I don't think every library sets this field in the same way.
There was a problem hiding this comment.
Hmm this test isn't expected to pass as of now, and exists as a result of extracting the tests that are expected to pass versus the ones that test fields that do not have a spec currently. I think keeping it here is fine right now, as it is marked as missing_feature due to lack of spec.
| assert link1.get("tracestate") == None | ||
|
|
||
|
|
||
| def _retrieve_span_links(span): |
There was a problem hiding this comment.
nit: can we move utils.parametric.spec.trace.retrieve_span_links() to a new shared module (ex: utils.trace.py) so that we can re-use this function in weblog tests.
There was a problem hiding this comment.
This was addressed by @cbeauchesne above. If a new weblog test will use the function, we can move it out then?
|
CI failures are the msgpack issue ? |
|
yes it is. Checking |
|
You may have found a real issue in this tracer, the agent does not like either the data from the lib, here is its response :
|
|
I've sent a message on slack to the java guild, let's wait for their response |
|
Waiting for #3587 to move forward. Once it's merged, we'll skip the test for spring-boot-native |
There was a problem hiding this comment.
Ok, almost good. There is still one weblog failing : https://github.com/DataDog/system-tests/actions/runs/12075461097/job/33675669767?pr=3499#step:25:57
|
The java error has been fixed by DataDog/dd-trace-java#8036 |
Motivation
Currently, we test the results of compound extraction of inconsistent headers through parametric tests. However, using the parametric app does not truly reflect the behavior exhibited through automatic instrumentation, as we have to manually call extract on the headers passed in, create a span, and add the extracted context/span links to the span. This PR aims to migrate the parametric tests into weblog tests to remove the manual process of imitating this scenario, and have the system-tests represent what actually happens when we have compound extraction on a set of inconsistent headers.
Changes
Add 2 weblog scenarios that take in
DD_TRACE_PROPAGATION_STYLE_EXTRACTwith values of tracecontext -> datadog -> b3 and datadog -> tracecontext -> b3 respectively. The tests create a span through themake_distant_callendpoint and we verify that we get back a span that has the appropriate span links for different scenarios.APMAPI-899
Workflow
codeownersfile quickly.🚀 Once your PR is reviewed, you can merge it!
🛟 #apm-shared-testing 🛟
Reviewer checklist
[<language>], double-check that only<language>is impacted by the changebuild-XXX-imagelabel is present