Skip to content

Conversation

@zanieb
Copy link
Member

@zanieb zanieb commented Jun 27, 2025

This fixes an obscure cache collision in Python interpreter queries, which we believe to be the root cause of CI flakes we've been seeing where a project environment is invalidated and recreated.

This work follows from the logs in this CI run which captured one of the flakes with tracing enabled. There, we can see that the project environment is invalidated because the Python interpreter in the environment has a different version than expected:

DEBUG Checking for Python environment at `.venv`
TRACE Cached interpreter info for Python 3.12.9, skipping probing: .venv/bin/python3
DEBUG The interpreter in the project environment has different version (3.12.9) than it was created with (3.9.21)

(this message is updated to reflect #14329)

The flow is roughly:

  • We create an environment with 3.12.9
  • We query the environment, and cache the interpreter version for .venv/bin/python
  • We create an environment for 3.9.12, replacing the existing one
  • We query the environment, and read the cached information

The Python cache entries are keyed by the absolute path to the interpreter, and rely on the modification time (ctime, nsec resolution) of the canonicalized path to determine if the cache entry should be invalidated. The key is a hex representation of a u64 sea hasher output — which is very unlikely to collide.

After an audit of the Python query caching logic, we determined that the most likely cause of a collision in cache entries is that the modification times of underlying interpreters are identical. This seems pretty feasible, especially if the file system does not support nanosecond precision — though it appears that the GitHub runners do support it.

The fix here is to include the canonicalized path in the cache key, which ensures we're looking at the modification time of the same underlying interpreter.

This will "invalidate" all existing interpreter cache entries but that's not a big deal.

This should also have the effect of reducing cache churn for interpreters in virtual environments. Now, when you change Python versions, we won't invalidate the previous cache entry so if you change back to the old version we can re-use our cached information.

It's a bit speculative, since we don't have a deterministic reproduction in CI, but this is the strongest candidate given the logs and should increase correctness regardless.

Closes #14160
Closes #13744
Closes #13745

Once it's confirmed the flakes are resolved, we should revert

@zanieb zanieb added the bug Something isn't working label Jun 27, 2025
@zanieb zanieb temporarily deployed to uv-test-registries June 27, 2025 20:23 — with GitHub Actions Inactive
@zanieb zanieb force-pushed the zb/canonical-interp branch from 383f8c2 to 3102646 Compare June 27, 2025 20:38
@zanieb zanieb temporarily deployed to uv-test-registries June 27, 2025 20:41 — with GitHub Actions Inactive
@zanieb zanieb marked this pull request as ready for review June 27, 2025 21:03
@zanieb zanieb requested a review from konstin June 27, 2025 23:22
@zanieb zanieb force-pushed the zb/canonical-interp branch from 3102646 to 3018210 Compare June 30, 2025 14:42
@zanieb zanieb temporarily deployed to uv-test-registries June 30, 2025 14:45 — with GitHub Actions Inactive
@zanieb zanieb enabled auto-merge (squash) June 30, 2025 15:33
@zanieb zanieb merged commit 1c7c174 into main Jun 30, 2025
162 of 165 checks passed
@zanieb zanieb deleted the zb/canonical-interp branch June 30, 2025 15:39
tmeijn pushed a commit to tmeijn/dotfiles that referenced this pull request Jul 6, 2025
This MR contains the following updates:

| Package | Update | Change |
|---|---|---|
| [astral-sh/uv](https://github.com/astral-sh/uv) | patch | `0.7.16` -> `0.7.19` |

MR created with the help of [el-capitano/tools/renovate-bot](https://gitlab.com/el-capitano/tools/renovate-bot).

**Proposed changes to behavior should be submitted there as MRs.**

---

### Release Notes

<details>
<summary>astral-sh/uv (astral-sh/uv)</summary>

### [`v0.7.19`](https://github.com/astral-sh/uv/blob/HEAD/CHANGELOG.md#0719)

[Compare Source](astral-sh/uv@0.7.18...0.7.19)

The **[uv build backend](https://docs.astral.sh/uv/concepts/build-backend/) is now stable**, and considered ready for production use.

The uv build backend is a great choice for pure Python projects. It has reasonable defaults, with the goal of requiring zero configuration for most users, but provides flexible configuration to accommodate most Python project structures. It integrates tightly with uv, to improve messaging and user experience. It validates project metadata and structures, preventing common mistakes. And, finally, it's very fast — `uv sync` on a new project (from `uv init`) is 10-30x faster than with other build backends.

To use uv as a build backend in an existing project, add `uv_build` to the `[build-system]` section in your `pyproject.toml`:

```toml
[build-system]
requires = ["uv_build>=0.7.19,<0.8.0"]
build-backend = "uv_build"
```

In a future release, it will replace `hatchling` as the default in `uv init`. As before, uv will remain compatible with all standards-compliant build backends.

##### Python

- Add PGO distributions of Python for aarch64 Linux, which are more optimized for better performance

See the [python-build-standalone release](https://github.com/astral-sh/python-build-standalone/releases/tag/20250702) for more details.

##### Enhancements

- Ignore Python patch version for `--universal` pip compile ([#&#8203;14405](astral-sh/uv#14405))
- Update the tilde version specifier warning to include more context ([#&#8203;14335](astral-sh/uv#14335))
- Clarify behavior and hint on tool install when no executables are available ([#&#8203;14423](astral-sh/uv#14423))

##### Bug fixes

- Make project and interpreter lock acquisition non-fatal ([#&#8203;14404](astral-sh/uv#14404))
- Includes `sys.prefix` in cached environment keys to avoid `--with` collisions across projects ([#&#8203;14403](astral-sh/uv#14403))

##### Documentation

- Add a migration guide from pip to uv projects ([#&#8203;12382](astral-sh/uv#12382))

### [`v0.7.18`](https://github.com/astral-sh/uv/blob/HEAD/CHANGELOG.md#0718)

[Compare Source](astral-sh/uv@0.7.17...0.7.18)

##### Python

- Added arm64 Windows Python 3.11, 3.12, 3.13, and 3.14

  These are not downloaded by default, since x86-64 Python has broader ecosystem support on Windows.
  However, they can be requested with `cpython-<version>-windows-aarch64`.

See the [python-build-standalone release](https://github.com/astral-sh/python-build-standalone/releases/tag/20250630) for more details.

##### Enhancements

- Keep track of retries in `ManagedPythonDownload::fetch_with_retry` ([#&#8203;14378](astral-sh/uv#14378))
- Reuse build (virtual) environments across resolution and installation ([#&#8203;14338](astral-sh/uv#14338))
- Improve trace message for cached Python interpreter query ([#&#8203;14328](astral-sh/uv#14328))
- Use parsed URLs for conflicting URL error message ([#&#8203;14380](astral-sh/uv#14380))

##### Preview features

- Ignore invalid build backend settings when not building ([#&#8203;14372](astral-sh/uv#14372))

##### Bug fixes

- Fix equals-star and tilde-equals with `python_version` and `python_full_version` ([#&#8203;14271](astral-sh/uv#14271))
- Include the canonical path in the interpreter query cache key ([#&#8203;14331](astral-sh/uv#14331))
- Only drop build directories on program exit ([#&#8203;14304](astral-sh/uv#14304))
- Error instead of panic on conflict between global and subcommand flags ([#&#8203;14368](astral-sh/uv#14368))
- Consistently normalize trailing slashes on URLs with no path segments ([#&#8203;14349](astral-sh/uv#14349))

##### Documentation

- Add instructions for publishing to JFrog's Artifactory ([#&#8203;14253](astral-sh/uv#14253))
- Edits to the build backend documentation ([#&#8203;14376](astral-sh/uv#14376))

### [`v0.7.17`](https://github.com/astral-sh/uv/blob/HEAD/CHANGELOG.md#0717)

[Compare Source](astral-sh/uv@0.7.16...0.7.17)

##### Bug fixes

- Apply build constraints when resolving `--with` dependencies ([#&#8203;14340](astral-sh/uv#14340))
- Drop trailing slashes when converting index URL from URL ([#&#8203;14346](astral-sh/uv#14346))
- Ignore `UV_PYTHON_CACHE_DIR` when empty ([#&#8203;14336](astral-sh/uv#14336))
- Fix error message ordering for `pyvenv.cfg` version conflict ([#&#8203;14329](astral-sh/uv#14329))

</details>

---

### Configuration

📅 **Schedule**: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).

🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied.

♻ **Rebasing**: Whenever MR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 **Ignore**: Close this MR and you won't be reminded about this update again.

---

 - [ ] <!-- rebase-check -->If you want to rebase/retry this MR, check this box

---

This MR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate).
<!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0MC42Mi4xIiwidXBkYXRlZEluVmVyIjoiNDAuNjIuMSIsInRhcmdldEJyYW5jaCI6Im1haW4iLCJsYWJlbHMiOlsiUmVub3ZhdGUgQm90Il19-->
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working

Projects

None yet

4 participants