fix(task): revert process group changes that cause hangs with nested mise tasks by jdx · Pull Request #8347 · jdx/mise

jdx · 2026-02-25T15:37:22Z

Summary

Reverts the setpgid/killpg changes from fix(task): use process groups to kill child process trees on Unix #8279 and fix(task): skip setpgid for TTY stdin to fix interactive tasks #8301
Placing each child process in its own process group via setpgid(0,0) breaks external tools (e.g. Playwright) that rely on process-group-based kills to tear down process trees
When Playwright SIGKILLs the webServer process group, grandchild processes in a separate group (created by inner mise's setpgid) survive, holding pipes open and causing the parent to hang indefinitely

Root Cause

The process group change created a hierarchy where nested mise run calls each placed their children in separate process groups:

mise run test  (pgid=terminal)
  └── sh -c "npm test"  (pgid=A, set by outer mise's setpgid)
      └── playwright
          └── mise run serve  (pgid=C, set by Playwright's detached:true)
              └── sh -c "npm run serve"  (pgid=D, set by inner mise's setpgid)
                  └── node main.js  (inherits pgid=D)

When Playwright kills pgid=C (kill(-pid, SIGKILL)), only processes in that group die. The actual server in pgid=D survives because SIGKILL cannot be caught/forwarded, and the surviving process holds pipe file descriptors open, preventing EOF and causing the parent mise to hang indefinitely.

Fixes #8345

Test plan

cargo build compiles cleanly
mise run lint passes
mise run test:unit — all 491 tests pass
mise run test:e2e test_task_timeout — timeout tests pass

🤖 Generated with Claude Code

Note

Medium Risk
Moderate risk because it changes how signals/timeouts propagate to subprocess trees on Unix, which can affect cleanup and Ctrl-C/termination semantics for running tasks.

Overview
Stops creating per-command process groups on Unix by removing the pre_exec setpgid hook, restoring the prior behavior where child commands stay in the caller’s process group.

Replaces all killpg usage with direct kill calls: timeout termination (SIGTERM then SIGKILL), global shutdown via CmdLineRunner::kill_all, and signal forwarding during execute now target only the child PID (and it no longer forwards SIGINT).

^{Written by Cursor Bugbot for commit e6ed47d. This will update automatically on new commits. Configure here.}

…mise tasks Reverts the setpgid/killpg changes from #8279 and #8301. Placing each child process in its own process group via setpgid(0,0) breaks external tools (e.g. Playwright) that rely on process-group-based kills to tear down process trees. When Playwright SIGKILLs the webServer process group, grandchild processes in a separate group (created by inner mise's setpgid) survive, holding pipes open and causing the parent to hang indefinitely. Fixes #8345 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

gemini-code-assist · 2026-02-25T15:38:38Z

Summary of Changes

Hello @jdx, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a critical issue where recent changes to process group management were causing hangs in nested mise tasks. By reverting these changes, the PR restores the previous behavior and resolves compatibility issues with tools like Playwright that rely on process group signaling.

Highlights

Reversion of Process Group Changes: This PR reverts the changes introduced in fix(task): use process groups to kill child process trees on Unix #8279 and fix(task): skip setpgid for TTY stdin to fix interactive tasks #8301 related to process group management.
Fix for Hanging Processes: The reverted changes were causing hangs with nested mise tasks due to issues with process group-based kills used by external tools like Playwright.
Root Cause Analysis: The process group changes created a hierarchy where nested mise run calls placed their children in separate process groups, leading to orphaned processes and hangs when signals were sent.

Changelog

src/cmd.rs
- Reverted process group changes to fix hangs with nested mise tasks.

Activity

The PR reverts changes from fix(task): use process groups to kill child process trees on Unix #8279 and fix(task): skip setpgid for TTY stdin to fix interactive tasks #8301.
The PR addresses issue Mise task hangs when a child processes is spawned with mise (since v2026.2.18) #8345.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request reverts process group management changes to fix issues with nested tasks. The changes primarily involve removing setpgid calls and replacing killpg with kill to target specific process IDs. The modifications appear correct and consistent with the goal of the pull request. I've identified one potential issue in the signal forwarding logic where an error from a kill call could be incorrectly propagated, and I've provided a suggestion to improve its robustness.

gemini-code-assist · 2026-02-25T15:39:30Z

src/cmd.rs

+                        debug!("Received signal {sig}, forwarding to {id}");
+                        let pid = nix::unistd::Pid::from_raw(id as i32);
+                        let sig = nix::sys::signal::Signal::try_from(sig).unwrap();
+                        nix::sys::signal::kill(pid, sig)?;


Using the ? operator here will cause execute() to return an error if kill fails. This can happen for benign reasons, such as the child process having already exited. This would incorrectly report a failure for the command execution.

To make this more robust and align with the behavior in TimeoutGuard, it's better to ignore the result of the kill call. The process we're trying to signal might already be gone, and that shouldn't be considered an error in this context.

Suggested change

nix::sys::signal::kill(pid, sig)?;

let _ = nix::sys::signal::kill(pid, sig);

greptile-apps · 2026-02-25T15:40:19Z

Greptile Summary

Reverted setpgid/killpg process group changes from PRs #8279 and #8301 to fix hangs with external tools like Playwright. The original changes placed each child process in its own process group, which broke tools that rely on process-group-based kills to tear down process trees. When Playwright killed its webServer process group with SIGKILL, grandchild processes in separate groups (created by nested mise calls) survived, held pipes open, and caused indefinite hangs.

Key changes:

Removed setpgid(0,0) call in pre_exec that created separate process groups
Changed back from killpg (process group kill) to kill (single process kill) in timeout handling and signal forwarding
Removed TTY detection logic that was conditionally applying process groups
Restored SIGINT filtering (skip forwarding SIGINT to avoid duplicate signals)

Trade-off: This revert sacrifices complete grandchild process cleanup (the original goal of #8279) to maintain compatibility with external tools that manage their own process groups. The issue with nested mise tasks and Playwright is prioritized over the grandchild orphaning problem.

Confidence Score: 4/5

This PR is safe to merge with minimal risk - it's a clean revert to previous working behavior
Score reflects that this is a well-understood revert with clear justification. The changes are internally consistent and return to previously stable code. The only concern is the known trade-off: grandchild processes may not be cleaned up as thoroughly (the original issue from before PR fix(task): use process groups to kill child process trees on Unix #8279), but this is an accepted compromise for Playwright compatibility
No files require special attention - the revert is clean and complete

Important Files Changed

Filename	Overview
src/cmd.rs	Reverts process group changes to fix Playwright/external tool hangs; returns to single-process kill approach

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["mise run test<br/>(terminal pgid)"] --> B["sh -c 'npm test'<br/>(pgid A - WITH setpgid)"]
    B --> C["playwright"]
    C --> D["mise run serve<br/>(pgid C - detached)"]
    D --> E["sh -c 'npm run serve'<br/>(pgid D - WITH setpgid)"]
    E --> F["node main.js<br/>(inherits pgid D)"]
    
    C -.->|"kill(-pid, SIGKILL)<br/>targets pgid C only"| G["PROBLEM: pgid D survives<br/>pipes stay open<br/>parent hangs"]
    
    F -.->|"survives because<br/>in different pgid"| G
    
    style G fill:#ff9999
    style F fill:#ffcccc
    style E fill:#ffcccc

_{Last reviewed commit: ca35256}

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}

src/cmd.rs

github-actions · 2026-02-25T15:57:47Z

Hyperfine Performance

`mise x -- echo`

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`mise-2026.2.20 x -- echo`	23.9 ± 0.4	23.4	28.0	1.02 ± 0.03
`mise x -- echo`	23.3 ± 0.5	22.7	30.0	1.00

`mise env`

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`mise-2026.2.20 env`	22.8 ± 0.6	22.1	29.6	1.00 ± 0.03
`mise env`	22.8 ± 0.3	22.1	24.8	1.00

`mise hook-env`

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`mise-2026.2.20 hook-env`	23.2 ± 0.4	22.7	27.7	1.00
`mise hook-env`	23.5 ± 0.5	22.8	29.2	1.01 ± 0.03

`mise ls`

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`mise-2026.2.20 ls`	20.9 ± 0.3	20.2	22.6	1.00
`mise ls`	21.2 ± 0.5	20.5	24.3	1.01 ± 0.03

`xtasks/test/perf`

Command	mise-2026.2.20	mise	Variance
install (cached)	149ms	150ms	+0%
ls (cached)	82ms	83ms	-1%
bin-paths (cached)	88ms	87ms	+1%
task-ls (cached)	825ms	823ms	+0%

…esses The child may have already exited (killed by TimeoutGuard or naturally) when a signal arrives, causing kill() to return ESRCH. Use `let _ =` instead of `?` to avoid spurious task failures from this race. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

@jdx

### 🐛 Bug Fixes - **(exec)** respect PATH order for virtualenv resolution in mise x by @jdx in [#8342](#8342) - **(task)** revert process group changes that cause hangs with nested mise tasks by @jdx in [#8347](#8347) - **(task)** resolve vars from subdirectory configs for monorepo tasks by @jdx in [#8343](#8343) - **(task)** resolve dependencies before prepare to fix monorepo glob deps by @jdx in [#8353](#8353) - python noarch with Conda backend by @wolfv in [#8349](#8349) ### New Contributors - @wolfv made their first contribution in [#8349](#8349) ## 📦 Aqua Registry Updates #### New Packages (3) - [`alexhallam/tv`](https://github.com/alexhallam/tv) - [`arcanist-sh/hx`](https://github.com/arcanist-sh/hx) - [`dathere/qsv`](https://github.com/dathere/qsv) #### Updated Packages (3) - [`astral-sh/ruff`](https://github.com/astral-sh/ruff) - [`caarlos0/fork-cleaner`](https://github.com/caarlos0/fork-cleaner) - [`rhysd/actionlint`](https://github.com/rhysd/actionlint)

…mise tasks (jdx#8347) ## Summary - Reverts the `setpgid`/`killpg` changes from jdx#8279 and jdx#8301 - Placing each child process in its own process group via `setpgid(0,0)` breaks external tools (e.g. Playwright) that rely on process-group-based kills to tear down process trees - When Playwright `SIGKILL`s the webServer process group, grandchild processes in a separate group (created by inner mise's `setpgid`) survive, holding pipes open and causing the parent to hang indefinitely ## Root Cause The process group change created a hierarchy where nested `mise run` calls each placed their children in separate process groups: ``` mise run test (pgid=terminal) └── sh -c "npm test" (pgid=A, set by outer mise's setpgid) └── playwright └── mise run serve (pgid=C, set by Playwright's detached:true) └── sh -c "npm run serve" (pgid=D, set by inner mise's setpgid) └── node main.js (inherits pgid=D) ``` When Playwright kills pgid=C (`kill(-pid, SIGKILL)`), only processes in that group die. The actual server in pgid=D survives because `SIGKILL` cannot be caught/forwarded, and the surviving process holds pipe file descriptors open, preventing EOF and causing the parent mise to hang indefinitely. Fixes jdx#8345 ## Test plan - [x] `cargo build` compiles cleanly - [x] `mise run lint` passes - [x] `mise run test:unit` — all 491 tests pass - [x] `mise run test:e2e test_task_timeout` — timeout tests pass 🤖 Generated with [Claude Code](https://claude.com/claude-code)  --- > [!NOTE] > **Medium Risk** > Moderate risk because it changes how signals/timeouts propagate to subprocess trees on Unix, which can affect cleanup and Ctrl-C/termination semantics for running tasks. > > **Overview** > Stops creating per-command process groups on Unix by removing the `pre_exec` `setpgid` hook, restoring the prior behavior where child commands stay in the caller’s process group. > > Replaces all `killpg` usage with direct `kill` calls: timeout termination (`SIGTERM` then `SIGKILL`), global shutdown via `CmdLineRunner::kill_all`, and signal forwarding during `execute` now target only the child PID (and it no longer forwards `SIGINT`). > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit e6ed47d. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup>  --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

@jdx

### 🐛 Bug Fixes - **(exec)** respect PATH order for virtualenv resolution in mise x by @jdx in [jdx#8342](jdx#8342) - **(task)** revert process group changes that cause hangs with nested mise tasks by @jdx in [jdx#8347](jdx#8347) - **(task)** resolve vars from subdirectory configs for monorepo tasks by @jdx in [jdx#8343](jdx#8343) - **(task)** resolve dependencies before prepare to fix monorepo glob deps by @jdx in [jdx#8353](jdx#8353) - python noarch with Conda backend by @wolfv in [jdx#8349](jdx#8349) ### New Contributors - @wolfv made their first contribution in [jdx#8349](jdx#8349) ## 📦 Aqua Registry Updates #### New Packages (3) - [`alexhallam/tv`](https://github.com/alexhallam/tv) - [`arcanist-sh/hx`](https://github.com/arcanist-sh/hx) - [`dathere/qsv`](https://github.com/dathere/qsv) #### Updated Packages (3) - [`astral-sh/ruff`](https://github.com/astral-sh/ruff) - [`caarlos0/fork-cleaner`](https://github.com/caarlos0/fork-cleaner) - [`rhysd/actionlint`](https://github.com/rhysd/actionlint)

gemini-code-assist bot reviewed Feb 25, 2026

View reviewed changes

cursor bot reviewed Feb 25, 2026

View reviewed changes

src/cmd.rs Outdated Show resolved Hide resolved

jdx merged commit 26ac453 into main Feb 25, 2026
48 of 51 checks passed

jdx deleted the fix/revert-process-groups branch February 25, 2026 20:00

mise-en-dev mentioned this pull request Feb 25, 2026

chore: release 2026.2.21 #8344

Merged

BrewTestBot mentioned this pull request Feb 26, 2026

mise 2026.2.21 Homebrew/homebrew-core#269534

Merged

jdx mentioned this pull request Mar 6, 2026

feat(tasks): minimal interactive runtime rework with segment rwlock #8473

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(task): revert process group changes that cause hangs with nested mise tasks#8347

fix(task): revert process group changes that cause hangs with nested mise tasks#8347
jdx merged 2 commits intomainfrom
fix/revert-process-groups

jdx commented Feb 25, 2026 •

edited by cursor bot

Loading

Uh oh!

gemini-code-assist bot commented Feb 25, 2026

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Feb 25, 2026

Uh oh!

greptile-apps bot commented Feb 25, 2026

Uh oh!

cursor bot left a comment

Uh oh!

Uh oh!

github-actions bot commented Feb 25, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

	nix::sys::signal::kill(pid, sig)?;
	let _ = nix::sys::signal::kill(pid, sig);

Uh oh!

Conversation

jdx commented Feb 25, 2026 • edited by cursor bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Root Cause

Test plan

Uh oh!

gemini-code-assist bot commented Feb 25, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Feb 25, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Feb 25, 2026

Greptile Summary

Confidence Score: 4/5

Important Files Changed

Flowchart

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Hyperfine Performance

mise x -- echo

mise env

mise hook-env

mise ls

xtasks/test/perf

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

jdx commented Feb 25, 2026 •

edited by cursor bot

Loading

github-actions bot commented Feb 25, 2026 •

edited

Loading

`mise x -- echo`

`mise env`

`mise hook-env`

`mise ls`

`xtasks/test/perf`