microsoft / microsoft/codecoverage
dotnet-coverage collect --server-mode leaks coverage across tests when using snapshot -r
Nobody has claimed this yet.
- Dominant language
- C#
- Stars
- 125
- Forks
- 17
- Avg merge
- 1h 17m
- Merged PRs (30d)
- 2
Description
We are observing that in a long-lived server-mode session, calling snapshot -r does not appear to fully clear prior coverage state. Coverage from an earlier test is still present in the next snapshot, and the raw .coverage file is already corrupted before any mapper/parsing logic executes. This is causing to collect the per test coverage when the test cases needs to run in sequence.
Steps:
- Start a shared coverage session:
cmd : dotnet-coverage collect --server-mode --background --session-id
2. Connect and run Test A in that same session:
cmd : dotnet-coverage connect dotnet test
3. Snapshot Test A with reset:
cmd : dotnet-coverage snapshot -r -o -t
dotnet-coverage merge -o -f xml
4. Connect and run Test B in the same session:
cmd : dotnet-coverage connect dotnet test
5. Snapshot Test B with reset:
cmd : dotnet-coverage snapshot -r -o -t
dotnet-coverage merge -o -f xml
6. Observe coverage contamination between Test A and Test B.
7. Shut down the server at the end
cmd: dotnet-coverage shutdown -t
Expected:
Each snapshot should produce coverage for only the test that executed since the previous reset. The -r option should clear the previous coverage state after creating the snapshot.
Actual:
The second snapshot still contains coverage from the earlier test or from a previous unrelated run. The raw .coverage output is already incorrect before mapper logic runs, which indicates the issue is in the coverage engine/server-mode reset behavior and not in downstream parsing.
Impact:
Incorrect coverage attribution, false pass/fail evidence, broken file-to-test mapping.
Note:
This issue is specific to dotnet-coverage collect --server-mode with repeated snapshot -r calls in the same process/session. It is not the same as MTP JSON-RPC server mode (--server / --server jsonrpc), which is a separate feature for IDE/tooling integration. The bug is therefore in the coverage collector/server-mode flow, not in the MTP test runner or mapper logic.
Relevant docs:
https://learn.microsoft.com/en-us/dotnet/core/testing/microsoft-testing-platform-code-coverage
https://learn.microsoft.com/en-us/dotnet/core/testing/microsoft-testing-platform-server-mode
https://github.com/microsoft/codecoverage
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the documented server-mode flow and reproduce the issue using the listed collect, connect, snapshot -r, merge, and shutdown commands for Test A and Test B. Inspect the raw snapshot outputs before merge or mapper processing; done means each snapshot contains only coverage collected since the preceding reset without corrupting the .coverage file.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- csharp
- Domain
- testing, tooling
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 52/100