"/usr/bin/kill: No such file or directory" in Docker.log when "Terminating com.docker.backend..."

• I have tried with the latest version of Docker Desktop
• I have tried disabling enabled experimental features
• I have uploaded Diagnostics
• Diagnostics ID: F37AEFC2-73B1-4F44-A83F-02FA5E1AC38C/20221121110657
• Diagnostics ID: (newer) F37AEFC2-73B1-4F44-A83F-02FA5E1AC38C/20221128003318

Expected behavior

Upon watchdog failure, Docker should be successfully terminated and then be relaunched.

Actual behavior

The docker VM kernel suffered some fault and Docker needed to terminate it.

"Docker is stopping...." hangs indefinitly. Docker.log indicates it is using /usr/bin/kill to kill process 900 (com.docker.backend), but there is "No such file or directory"

time="2022-10-24T03:28:04Z" level="info" msg="GoBackend: Terminating com.docker.backend (89006) and waiting until exit"
time="2022-10-24T03:28:04Z" level="info" msg="GoBackend: Closing com.docker.backend wathdog pipe"
time="2022-10-24T03:28:04Z" level="info" msg="GoBackend: Waiting for com.docker.backend to finish."
time="2022-10-24T03:29:44Z" level="info" msg="GoBackend: com.docker.backend (89006) has still not exited, killing it."
time="2022-10-24T03:29:44Z" level="error" msg="Cmd: /bin/sh: /usr/bin/kill: No such file or directory
"

On my macOS /usr/bin/kill indeed does not exist; instead /bin/kill does. Using kill -QUIT 900 immediately recovered the situation and I was able to successfully restart Docker.

Information

You can see the path is present in the Docker binary:

$ strings /Applications/Docker.app/Contents/MacOS/Docker | grep /usr/bin/kill
/usr/bin/kill -9

Reproducible: perhaps, if you could cause the VM kernel to become unresponsive. I did notice that a 'node' process (presumably from VSCode Dev Container) was timing out (hung kernel task) buried within do_statx. I had been out and about working remotely for part of the day, but had been working happily for a couple of hours by the time it crashed.

New Problem?: I think I noticed increased instability for a few updates now.... perhaps a few months.

• macOS Version: Monterey 12.6.1
• Intel chip or Apple chip: Apple M1 Pro
• Docker Desktop Version: 4.13.1 (90346)

Output of /Applications/Docker.app/Contents/MacOS/com.docker.diagnose check

(no longer relevant, but at the the time it did detect that Docker wasn't running, and reported numerous EOF errors)

Issue recurred; uploaded new diagnostics report.

I was using VS Code with Remote Containers at the time, not doing anything particular, just editing a file, when the connection to Docker was reported to have failed by VS Code.

The following was again observed in Docker.log

time="2022-11-28T00:33:23Z" level="info" msg="GoBackend: com.docker.backend (76220) has still not exited, killing it."
time="2022-11-28T00:34:01Z" level="error" msg="Cmd: /bin/sh: /usr/bin/kill: No such file or directory

Earliest sign of error is in ./log/host/com.docker.driver.amd64-linux.log

[2022-11-28T00:30:47.691025000Z][com.docker.driver.amd64-linux][I] reading from vpnkit: EOF
[2022-11-28T00:30:47.696229000Z][com.docker.driver.amd64-linux][W] Error closing raw stream to client: close unix <HOME>/.docker/run/docker.sock->: shutdown: socket is not connected

Last time this happened it happened on a Monday also.

I am now resetting (Docker) to factory defaults... although I had to kill the backend process to be able to do this from the GUI.

Noticed there is an update available: updating from 4.13.1 (90346) to 4.14.1 (91661).

[nicks]

thanks for the report! we're investigating.

To clarify, has this been actually fixed (if so, should I expect it in the next release?) or has it been closed for some other reason?

The problem is still there in 4.14.1

$ strings /Applications/Docker.app/Contents/MacOS/Docker | grep /usr/bin/kill
/usr/bin/kill -9

Thanks,
Cameron

[mat007]

A fix has been released in Docker Desktop 4.16. See the release notes for more details.

[docker-robott]

Closed issues are locked after 30 days of inactivity.
This helps our team focus on active issues.

If you have found a problem that seems similar to this, please open a new issue.

/lifecycle locked