Bun development tools fail when process exits

“What version of Bun is running? 1.4.3+c6da4a4d3 (also reproduces on 1.4.4 canary.1+5f5799ed1) What platform is your computer? Linux 7.0.0 38 generic x86 64 unknown What steps can reproduce the bug? When the shell for a package.json script cannot be spawned, bun run prints an error but exits with status 0 . A script whose command line exceeds ARG MAX triggers it reliably ( posix spawn → E2BIG ): Output (same on 1.4.3 and 1.4.4 canary.1): stderr for bun run big ends with: What is the expected beha”

“What version of Bun is running? 1.4.3+c6da4a4d3, and 1.4.4 canary.1+6c1eb067e What platform is your computer? Linux 7.2.8 arch1 2 x86 64 What steps can reproduce the bug? What is the expected behavior? server exit code: null : the server is still listening, so its process keeps running whatever inspector clients come and go. What do you see instead? beforeExit fires in the server and the process exits with code 0 while Bun.serve is still listening, as if the event loop stopped counting the serve”

“What version of Bun is running? 1.4.3+c6da4a4d3 (stable); 1.4.4 canary.1+6c1eb067e behaves the same What platform is your computer? Linux 7.2.9 (WSL2) x86 64 What steps can reproduce the bug? What is the expected behavior? The same as Node (v26.11.1): ENOENT . With an empty PATH , ls is not found. What do you see instead? Other variants on Bun 1.4.3, from the same script with the line that sets the environment changed: | Setup | Node | Bun | | | | | | process.env.PATH = "" | ENOENT | ran | | pro”

“What version of Bun is running? 1.4.2+744846f84. Also reproduces on 1.4.3 canary.1+5749c3129. What platform is your computer? Linux 7.2.6 x86 64. Also seen on GitHub Actions ubuntu 22.04 runners. What steps can reproduce the bug? This is the full packages/ai suite (483 test files) of oh my pi, at a commit on our public fork RigelBuild/oh my pi. I don't have a small repro. The hang depends on how many files run in one process, and no single file triggers it. parallel implies isolate . With parall”

“What version of Bun is running? 1.4.3 canary.1+5f554969b What platform is your computer? Linux 7.0.0 31 generic x86 64 unknown What steps can reproduce the bug? A busy Worker keeps executing JavaScript for tens to hundreds of milliseconds after worker.terminate() is called. The same program on Node stops the worker within a few milliseconds. spin.js : the worker spins forever, stamping the time it last ran into shared memory: repro.js : measures how long the worker keeps running after terminate(”

“What version of Bun is running? 1.4.2+744846f84 What platform is your computer? Microsoft Windows NT 10.0.26200.0 x64 What steps can reproduce the bug? a simple vite dev server when terminated with "Ctrl + C" in windows, next run command gives the error the port is already in use and i can see the process still being alive. What is the expected behavior? process must be terminated What do you see instead? No response Additional information No response”

“What version of Bun is running? 1.3.14 (aarch64 apple darwin), also seen on 1.3.13 What platform is your computer? macOS 15.1, arm64 (M series). Also reproduces on GitHub's macos 14 runners. What steps can reproduce the bug? Spawn a PTY child via Bun.spawn(..., { terminal }) that exits on its own, and race a process group SIGTERM / SIGKILL against that exit. Within a few thousand iterations the process stops making progress entirely. What is the expected behavior? The loop runs to completion. Ra”

“What version of Bun is running? 1.4.2+744846f84 (also reproduced on 1.4.0 and 1.4.1) What platform is your computer? Microsoft Windows NT 10.0.26200.0 x64 What steps can reproduce the bug? repro.ts : The same happens with a file redirect ( bun run repro.ts < in.txt ) and with a parent that spawns the script with stdin: "pipe" and writes to it two seconds later — the child exits before the data arrives. What is the expected behavior? This is what Linux prints on 1.4.0, and what Windows prints on”

“What version of Bun is running? Bun 1.4.0 on Windows x64. What platform is your computer? Windows 11 x64. What steps can reproduce the bug? Create target.js : Create replace.js : In terminal 1: In terminal 2: On Windows this fails consistently with an error like: The same behavior occurs with: Controls tested on the same machine: bun target.js + the same rename(temp, target) succeeds. node watch target.js + the same atomic replacement succeeds (Node v26.8.1). Directly overwriting/truncating targ”

“What version of Bun is running? 1.4.0 (stable), macOS arm64 (Darwin 25.5.0) What steps can reproduce the bug? No minimal repro yet — filing with live captured evidence because the failure leaves no trace otherwise. Real world shape that reproduces at 3/8 full runs: bun test parallel=12 (implies isolate ) over a 205 file suite in a Bun + Turborepo monorepo, with turbo building/testing sibling packages concurrently (high host load at worker startup). Each test file's setup awaits one memoized init”

“What version of Bun is running? 1.3.14+0d9b296af (also reproduced on canary 1.4.0) What platform is your computer? Darwin 24.6.0 arm64 arm, Linux 6.18.9 talos x86 64 What steps can reproduce the bug? Under Bun, when the connection is TLS ( wss:// ) and the network path is dead (packets are silently dropped), terminate() moves the socket to readyState 2 (CLOSING) and the close event never fires. The socket stays in CLOSING indefinitely. Any code waiting on close to react to the termination hangs”

“What version of Bun is running? 1.4.0 canary.1 What platform is your computer? x64 What steps can reproduce the bug? // debugId=176A91D20735B3A564756E2164756E21 // sourceMappingURL=/chunk yhrkxxeb.js.map and those urls are exposed by the server. What is the expected behavior? it is expected that users should be able to enable or disable sourcemaps for production mode Bun.serve What do you see instead? I can see sourcemap in compiled chunks and i can access those sourcemaps from the server. Addit”

“What version of Bun is running? 1.3.14+0d9b296af What platform is your computer? Linux 7.0.0 27 generic x86 64 unknown What steps can reproduce the bug? If I have a Bun script that uses a timer, and then I put my machine into suspend, and then later resume the machine, the timer will not continue. It doesn't matter which method I use to implement the timer all these three implementations exhibit the issue: setInterval: setTimeout: Bun.sleep What is the expected behavior? After I resume the machi”

“What version of Bun is running? 1.4.0 canary.1+98094dc4d What platform is your computer? Darwin 24.6.0 arm64 arm What steps can reproduce the bug? Calling process.exit() in watch mode terminates Bun entirely including the watcher itself. This makes it difficult to develop and test CLIs & scripts with Bun where you need to work with cases like this since I'd have to manually restart every time. For example: What is the expected behavior? The watcher should remain active when the process exists an”

“What version of Bun is running? 1.3.14+0d9b296af What platform is your computer? Linux 6.8.0 124 generic x86 64 x86 64 What steps can reproduce the bug? create file test.js and file ./helpers/script.js then run bun test.js What is the expected behavior? Node.js outputs What do you see instead? Bun outputs which is the opposite order Additional information No response”

“What version of Bun is running? 1.4.0 canary.1+7069b3092 (also reproduces on 1.3.14+0d9b296af ) What platform is your computer? Microsoft Windows NT 10.0.26200.0 x64 What steps can reproduce the bug? Complete, runnable reproduction: https://github.com/BlackHole1/bun detached windows bug ( pwsh run.ps1 runs it and reports whether each spawned helper survived the parent's exit; includes a passing cmd /c start workaround variant and a multi run stability check.) The bug in one file — on Windows, a”

“What version of Bun is running? 1.3.14 canary.1+d5945cffa What platform is your computer? Linux 6.12.86+deb13 amd64 x86 64 x86 64 What steps can reproduce the bug? Hi, I would like to report a behavior in Bun that can be reproduced using the below snippet: What is the expected behavior? HELLO WORLD What do you see instead? Additional information No response”

“What version of Bun is running? 1.3.11 canary.1+9e93bfa1b What platform is your computer? Linux 5.10.0 12 amd64 x86 64 x86 64 What steps can reproduce the bug? Hi, I would like to report a behavior in Bun that can be reproduced using the below PoC: 1. worker.mjs 2. foo.mjs 3. package.json 4. npm install 5. bun run foo.mjs What is the expected behavior? Print 10 and exit What do you see instead? Prints 10 and does not finish execution Additional information No response”

“What version of Bun is running? 1.3.7 canary.49+f833f11af What platform is your computer? Linux 5.15.0 164 generic x86 64 x86 64 What steps can reproduce the bug? What is the expected behavior? What do you see instead? Additional information MySQL server version: 8.0.44 0ubuntu0.22.04.2 OS: Ubuntu 22.04.5 LTS Mysql2 has no problems with auth using caching sha2 password , and I can connect with the same credential via phpMyAdmin and mysql CLI as well. If I change the auth plugin to mysql native p”

“What version of Bun is running? 1.3.0 What platform is your computer? Microsoft Windows NT 10.0.26200.0 x64 What steps can reproduce the bug? 1. Install dependencies: 2. Create a test file: 3. Run with Bun: What is the expected behavior? Should read barcodes from the image and print results, or throw an error before the last line executes. What do you see instead? Bun exits silently after "Read image data from..." log. No error is thrown, the process just terminates. This happens non determinist”

“What version of Bun is running? 1.2.19+aad3abead What platform is your computer? Microsoft Windows NT 10.0.26100.0 x64 What steps can reproduce the bug? Discrovered in a monorepo project using Bun with a structure like this: I use the following command from the root package.json: Steps to reproduce: 1. Clone a Bun monorepo project with separate client/ and server/ folders. 2. Add the above root script that uses bun cwd <dir run dev inside concurrently. 3. Start with bun dev . 4. Press Ctrl+C. 5.”

“What version of Bun is running? 1.1.38 What platform is your computer? x64 What steps can reproduce the bug? Very simple code, it works fine in nodejs, but it doesn't run in bunjs. It doesn't report any errors, but it gets stuck and there is no output. Image What is the expected behavior? Run it normally What do you see instead? No output, always stuck Additional information No response”

“What version of Bun is running? 1.1.26+0a37423ba What platform is your computer? Linux 6.8.0 40 generic x86 64 x86 64 What steps can reproduce the bug? 1. Input this code into bun cluster.mjs file: 2. Run the code in watch mode: 3. Press Ctrl+C and see that process still work.”

“What version of Bun is running? 1.1.10+5102a9443 What platform is your computer? Microsoft Windows NT 10.0.19045.0 x64 What steps can reproduce the bug? main.ts package.json What is the expected behavior? The following printed to stdout on interrupt (Ctrl+C) when either of the following commands is ran: bun main.ts bun start What do you see instead? When running bun main.ts , the output is correct. When running bun start , the output is either empty or truncated. Additional information Invoking”

“What version of Bun is running? 1.0.2+37edd5a6e389265738e89265bcbdf2999cb81a49 What platform is your computer? Darwin 22.5.0 arm64 arm What steps can reproduce the bug? I wrote a custom loader (plugin) to support importing .scss files. The loader is working as expected but when using bun watch , the imported .scss file does not rerun the process on save. You can reproduce by doing the following: Create a .scss file styles.scss Create a plugin sassLoader.js (and preload the plugin in bunfig.toml”