james-willis commented on PR #1128: URL: https://github.com/apache/sedona-db/pull/1128#issuecomment-5209065227
Verification note, since a `context.rs`-only change doesn't match the `python-wheels` `paths:` filter and so isn't exercised by this PR's own CI: The crash was reproduced locally on a free-threaded CPython 3.14 build (`python-freethreading`) running the full test suite — it fails partway through with `OS can't spawn worker thread` (EAGAIN), matching the CI signature, and only on the free-threaded interpreter (every GIL build passes the same tests). Instrumenting the existing `RuntimeHandle` teardown ruled out a hung or leaking janitor: every runtime drained cleanly (`drops == drained`). The cause was the *number* of per-session runtimes alive at once — dozens draining concurrently, each with a ~`num_cpus` worker pool. With the single shared runtime the same suite goes from crashing to **zero spawn failures**, with at most one runtime ever outstanding. A full `python-wheels` run is being used separately to confirm the `cp314t` macOS wheel builds and tests clean end-to-end: https://github.com/james-willis/sedona-db/actions/runs/31127918558 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
