Warn at probe time on SDK/worker Python minor mismatch#60
Merged
Conversation
probe_endpoint_worker (used by `uxarray-mcp doctor`) now compares the submitter's Python minor version to the worker's reported version. When they differ, the probe payload gains a warning explaining that high-level MCP tools survive the skew via AllCodeStrategies but raw Executor.submit() calls will hit pickle protocol errors and WorkerLost. The single existing PYTHONPATH warning is preserved for back-compat under payload["warning"] and the new accumulator also exposes payload["warnings"] as a list for clients that want both. Docs: docs/operating-an-endpoint.md gains a callout in the solo personal quickstart explaining the same constraint and pointing at the doctor warning. The example conda env is also bumped from python=3.11 to python=3.12 to match what most HPC sites ship today. Rationale: tripped on this myself today while running an inline Executor.submit(_ls, ...) against chrysalis from a 3.13 laptop against a 3.12 worker. Surfacing it at probe time would have saved 30 minutes of diagnosis.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Why
Hit this myself on chrysalis today: high-level MCP tools (`inspect_mesh_remote`, `calculate_area_remote`) survived a 3.13 → 3.12 SDK/worker skew because they route through `AllCodeStrategies`, but a raw `Executor.submit()` from a one-off script blew up with `WorkerLost`. The error message Globus Compute prints is generic enough that it took 30 minutes to recognize the root cause. Surfacing the mismatch at probe time means the next person sees the fix command directly.
Test plan