Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

`reflex run --json` ignores SIGINT sent to its pid; under supervisord stopsignal=INT the app is orphaned and blocks the restart

Open Beginner friendly
  • Is something wrong?
#7,509 1 comment 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

@harsh21234i is already working on this.

Since Oct 14, 2026.

  • #7558 by @harsh21234i — open

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
76/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
python
Domain
backend

Research direction

Start in packages/reflex-base/src/reflex_base/utils/log.py, especially the SIGTERM forwarding at line 519 and the KeyboardInterrupt handling at lines 520–526. Check how the supervising process waits for and signals its child, then reproduce with reflex run --json and SIGINT sent to its pid. Done means the child exits promptly and the server does not remain running; rerun the issue’s repro matrix to check the behavior.

Written by the indexing model from the issue text.

Description

Severity: LOW, pre-existing. 0.10.0a2 behaves like a3; 0.9.12 also needs a SIGKILL, but there nothing survives it. Confirmed by an independent verifier (pre-release finding A3-07).

Symptom. kill -INT <pid> of reflex run --json does nothing: the server is still serving 12–210 s later. Every other way of stopping it works quickly:

stop method time to exit
plain reflex run, SIGINT 0.18–0.25 s
--json, SIGTERM to the pid 0.29 s–2 s
--json, SIGINT to the process group (Ctrl-C) 0.27 s
--json, SIGINT to every pid in the tree (systemd's default KillMode=control-group) 0.25 s

Cause. The supervising process catches SIGINT, polls wait4(WNOHANG) for 0.25 s and goes back to a blocking wait without signalling the child. packages/reflex-base/src/reflex_base/utils/log.py:519 forwards only SIGTERM, and :520-526 does except KeyboardInterrupt: continue.

Under supervisord stopsignal=INT (default stopasgroup/killasgroup):

  • the stop waits 10.2 s, then SIGKILLs only the supervisor pid;
  • the inner reflex, the backend worker, bun and node become orphans that keep serving;
  • the next supervisorctl start fails 4× with "Frontend port: 3640 is already in use" and goes FATAL.

With stopasgroup=true it stops in 0.42 s, and with the default TERM config in 0.59 s.

Also affected:

  • systemd with KillMode=mixed/process and KillSignal=SIGINT;
  • proc.send_signal(SIGINT);
  • docker --init (tini) or running as PID 1 (simulated: hung 15 s);
  • timeout --foreground -s INT.

GNU timeout -s INT signals the whole group, so it isn't affected. This contradicts the 0.10.0a1 changelog entry for #7328 ("stops its frontend on SIGTERM and SIGINT without a TTY") for --json.

Fix. Forward SIGINT to the child the same way SIGTERM is forwarded in log.py.

Repro: a3_upgrade/bin/json_matrix.sh a3fast a3 INT-pid:100000 (dir). Evidence: A3-07 in FINDINGS.md.

Dominant language
Python
Stars
28.9k
Forks
1.8k
Avg merge
2d 14h
Merged PRs (30d)
211

Getting set up

Open in Codespaces

Starts the project's dev container in your browser, under your own GitHub account.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from reflex-dev/reflex

All issues in reflex-dev/reflex

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.