Drop unsampled transactions when connected to APM Server 8.0+

elastic / apm

Elastic Application Performance Monitoring - resources and general issue tracking for Elastic APM.

https://www.elastic.co/apm

Apache License 2.0

384 stars 114 forks source link

Drop unsampled transactions when connected to APM Server 8.0+ #554

Closed felixbarny closed 2 years ago

felixbarny commented 3 years ago

May the instrumentation collect sensitive information, such as secrets or PII (ex. in headers)?
- [ ] Yes
- [ ] Add a section to the spec how agents should apply sanitization (such as sanitize_field_names)
- [ ] No
- [ ] Why?
- [x] n/a
[x] Link to meta issue: #545
[x] Create PR as draft
[x] Approval by at least one other agent
[x] Mark as Ready for Review (automatically requests reviews from all agents and PM via CODEOWNERS)
- Remove PM from reviewers if impact on product is negligible
- Remove agents from reviewers if the change is not relevant for them
[x] Approved by at least 2 agents + PM (if relevant)
[ ] Merge after 7 days passed without objections
[x] Create implementation issues (ideally add a milestone)
[x] Create a status table and add it to the meta issue

/schedule 2021-12-08

apmmachine commented 3 years ago

:green_heart: Build Succeeded

the below badges are clickable and redirect to their specific view in the CI or DOCS

Expand to view the summary

#### Build stats * Start Time: 2021-12-06T05:08:00.601+0000 * Duration: 4 min 31 sec * Commit: cce0139b7c505b1fdc1f552d414befafffd2cfad

felixbarny commented 2 years ago

This is scheduled to be merged by tomorrow EOB (see the /schedule command in the PR description).

beniwohli commented 2 years ago

A questions that came up while implementing this:

For distributed tracing, we currently propagate the headers even for unsampled transactions, using the transaction.id as the parent-id. This doesn't seem to make much sense if there is no matching transaction document recorded. Should we stop propagating the traceparent headers for unsampled transactions?

axw commented 2 years ago

@beniwohli that would lead to the downstream service starting a new trace right? That doesn't seem desirable.

felixbarny commented 2 years ago

To add to Andrew's point, even if we're not recording and storing transactions, generating and propagating the IDs is still relevant for log correlation.