podman-remote run -it: failed to resize TTY: container in wrong state "stopped|created"

[edsantiago]

In window 1:

$ ./bin/podman system service --timeout=0

In window 2:

$  ./bin/podman-remote run --rm -it quay.io/libpod/alpine echo hello
hello
WARN[0000] failed to resize TTY: container "87e8cefa1dd42c9a819b76ae758d12a3e2358696320e0f178e0db3deabbb1195" in wrong state "stopped"
WARN[0000] Container 87e8cefa1dd42c9a819b76ae758d12a3e2358696320e0f178e0db3deabbb1195 does not exist: no container with name or ID "87e8cefa1dd42c9a819b76ae758d12a3e2358696320e0f178e0db3deabbb1195" found: no such container

The errors are not deterministic. Sometimes the state is "created" instead of "stopped". Sometimes the second message is an error instead of warning:

$  ./bin/podman-remote run --rm -it quay.io/libpod/alpine echo hello
hello
WARN[0000] failed to resize TTY: container "49980f36f6f2c0adddcdc4949199c2b58020074322617dad897e3188d7756f2c" in wrong state "stopped"
ERRO[0000] Error removing container 49980f36f6f2c0adddcdc4949199c2b58020074322617dad897e3188d7756f2c: error looking up container "49980f36f6f2c0adddcdc4949199c2b58020074322617dad897e3188d7756f2c" mounts: layer not known

(Even though it says ERROR, exit code is always 0).

Sometimes there's no second line at all, only the first.

master @ 9e23e0b

[edsantiago]

Forgot to mention: there's a simple workaround, just add a sleep:

$ ./bin/podman-remote run --rm -it quay.io/libpod/alpine sh -c 'sleep 0.3;echo hello'
hello

0.2 is not quite good enough, and 0.1 is even less good.

[edsantiago]

For reference: flake failure in CI