Non-terminal signals are killing containers running with --init flag

[rayrapetyan]

Description
This looks like invalid behavior: docker-init (tini) process should proxy signals to child processes.
Instead any container started with --init flag can be killed with a non-terminal signal (e.g. SIGUSR2, SIGHUP etc).

Reproduce

1. docker run --init --rm -it -p 80:80 --name hello strm/helloworld-http
2. docker kill --signal SIGUSR2 hello

Expected behavior
Container continues to run (as it does when started without --init flag)

Actual
Container is being stopped and killed

docker version

Client: Docker Engine - Community
 Version:           20.10.21
 API version:       1.41
 Go version:        go1.18.7
 Git commit:        baeda1f
 Built:             Tue Oct 25 18:02:28 2022
 OS/Arch:           linux/amd64
 Context:           default
 Experimental:      true

Server: Docker Engine - Community
 Engine:
  Version:          20.10.21
  API version:      1.41 (minimum version 1.12)
  Go version:       go1.18.7
  Git commit:       3056208
  Built:            Tue Oct 25 18:00:19 2022
  OS/Arch:          linux/amd64
  Experimental:     false
 containerd:
  Version:          1.6.10
  GitCommit:        770bd0108c32f3fb5c73ae1264f7e503fe7b2661
 runc:
  Version:          1.1.4
  GitCommit:        v1.1.4-0-g5fd4c4d
 docker-init:
  Version:          0.19.0
  GitCommit:        de40ad0

docker info

Client:
 Context:    default
 Debug Mode: false
 Plugins:
  app: Docker App (Docker Inc., v0.9.1-beta3)
  buildx: Docker Buildx (Docker Inc., v0.9.1-docker)
  compose: Docker Compose (Docker Inc., v2.12.2)
  scan: Docker Scan (Docker Inc., v0.21.0)

Server:
 Containers: 1
  Running: 0
  Paused: 0
  Stopped: 1
 Images: 32
 Server Version: 20.10.21
 Storage Driver: overlay2
  Backing Filesystem: extfs
  Supports d_type: true
  Native Overlay Diff: true
  userxattr: false
 Logging Driver: json-file
 Cgroup Driver: systemd
 Cgroup Version: 2
 Plugins:
  Volume: local
  Network: bridge host ipvlan macvlan null overlay
  Log: awslogs fluentd gcplogs gelf journald json-file local logentries splunk syslog
 Swarm: inactive
 Runtimes: io.containerd.runc.v2 io.containerd.runtime.v1.linux nvidia runc
 Default Runtime: runc
 Init Binary: docker-init
 containerd version: 770bd0108c32f3fb5c73ae1264f7e503fe7b2661
 runc version: v1.1.4-0-g5fd4c4d
 init version: de40ad0
 Security Options:
  apparmor
  seccomp
   Profile: default
  cgroupns
 Kernel Version: 5.10.0-19-amd64
 Operating System: Debian GNU/Linux 11 (bullseye)
 OSType: linux
 Architecture: x86_64
 CPUs: 16
 Total Memory: 31.14GiB
 Name: ara-home-pc-debian
 ID: 3KXM:A6UU:LLYL:UYPH:ZWH2:AJ37:G5UL:2UZC:653T:BFLZ:R5OD:DOSR
 Docker Root Dir: /var/lib/docker
 Debug Mode: false
 Registry: https://index.docker.io/v1/
 Labels:
 Experimental: false
 Insecure Registries:
  127.0.0.0/8
 Live Restore Enabled: false

[tianon]

I think this means your process doesn't have a handler for SIGUSR2 registered; from signal(7) (https://manpages.debian.org/bullseye/manpages/signal.7.en.html#Standard_signals), SIGUSR2 is a terminating signal by default.

[rayrapetyan]

Hm, I didn't know SIGUSRs are terminating by default, that's why used a random web server as example, sorry. But initially this issue happened with my own image which has a SIGUSR2 handler. I've created a POC repo: https://github.com/rayrapetyan/test_sig.

Steps to reproduce the issue:

1. Run cmd.sh locally
2. See output:

PID of this script: 433226
Running....
Running....
...

3. Send SIGUSR2 signal: kill -SIGUSR2 433226
4. Notice signal was intercepted in cmd.sh:

handling SIGUSR2
Running....

and process continues running as expected.

Now run cmd.sh from container:

docker build -t test_sig:latest .
docker run --rm --init --name test_sig test_sig:latest

Send SIGUSR2:
docker kill --signal SIGUSR2 test_sig

Expected: cmd.sh running inside a docker container should receive and handle a signal and continue running.
Actual: container stops.

Now after looking into how cmd.sh is running inside a container:
/sbin/docker-init -- /bin/sh -c ./cmd.sh -> bash cmd.sh
I realize that because of /bin/sh is running before bash, it's intercepting all signals and doesn't forward them to a child processes (bash) properly?

[rayrapetyan]

Works when running as:
docker run --rm --init --name test_sig test_sig:latest ./cmd.sh
so no /bin/sh in between.