title: "pipeio_run_kill: "Failed to kill screen session" despite the screen existing
"¶
status: open created: 2026-04-17 updated: 2026-04-17 timestamp: 20260417-194148-680260 tags: [issue] source: agent-observation project_primary: projio capture_id: 20260417-194147-746991 confidence: 1.0 transcript_file: /storage2/arash/worklog/workflow/captures/20260417-194147-746991/transcript.txt
pipeio_run_kill: "Failed to kill screen session" despite the screen existing
Observation¶
During this session, pipeio_run_kill(run_id="1a4b71a9") returned:
{"error":"Failed to kill screen session 'pipeio-preprocess_ieeg-1a4b71a9'"}
even though screen -ls showed the session was still running. Manual fallback: ps aux | grep snakemake | kill -TERM.
Expected behavior¶
Either kill the screen session reliably (via screen -S <name> -X quit) or report precisely why it couldn't (session not found, permission denied, etc).
Impact¶
Had to drop to Bash and manually kill processes, which the user had asked me to avoid in favor of pipeio tools. This contributed to a ~5-minute detour and required the user to remind me to use MCP tools.
Suggested investigation¶
- Does pipeio_run_kill use
screen -X quitor signals to the child PID? - When the MCP returns "failed to kill", is the screen actually killed or left orphaned?
- Would a fallback chain (screen quit → SIGTERM → SIGKILL on child tree) be more reliable?
Source context: pixecog¶
PixEcog (pixecog): Neuropixels and ECoG dataset and analysis
Recent commits:
bee27f8 preprocess_ieeg: clock correction + mod refactor + two-pass bad channel detection
36d420f preprocess_ieeg: clock correction + mod refactor + two-pass badchannel + linenoise aliased harmonics
61a7fe8 [DATALAD] Added subdataset
README:
type: readme
Quick Start for Collaborators¶
Follow this checklist to get started with Pixecog documentation and workflows.
🐀 Pixecog Project — Compact Overview¶
Core principles
- One immutable BIDS raw dataset (
raw/) as the canonical baseline - Each analysis pipeline ha
Related Notes¶
- issue-arash-20260417-194119-134127.md — Both are pipeio_run MCP tool bugs observed in the same session
- issue-arash-20260414-030010-945964.md — Another pipeio_run execution bug (env resolution)
- issue-arash-20260412-013511-941571.md — Test coverage for pipeio package — kill/session management would be a gap to cover
- issue-arash-20260409-232128-727233.md — pipeio_run snakemake invocation bug — same tool, execution reliability theme