Release v2.0.22 fails to run under Ubuntu 18.04

[ksdd]

Release v2.0.22 fails to run under Ubuntu 18.04 with the following error message

Error: failed to connect to container's attach socket: /run/user/205013812/libpod/tmp/socket/7c19123d148494b91f6689e0882d9f2b1b7b7bb8036128bbd4c363c11bbe3044/atta: dial unixpacket /run/user/205013812/libpod/tmp/socket/7c19123d148494b91f6689e0882d9f2b1b7b7bb8036128bbd4c363c11bbe3044/atta: connect: no such file or directory

I tested the binary provided by the recommended repo as well as the official release binary. Both fail to run with the same error.

BTW: When running the official release binary with --version option, I get:

$ /usr/libexec/podman/conmon --version
conmon version 2.0.23-dev
commit: 715c680-dirty

Obviously the binary has been build from a dirty working copy and is NOT based on the commit 9c34a86 tagged as v2.0.22!

On the other hand the binary from the recommended repo seems to be build based on v2.0.22 according to its version output.

[haircommander]

how are you running conmon? is this with podman or CRI-O?

[ksdd]

I run it via podman.

[haircommander]

what was the podman command you used? and what was the podman version?

[ksdd]

$ podman --version
podman version 2.2.1

$ podman run --interactive --tty --rm alpine
Error: failed to connect to container's attach socket: /run/user/205013812/libpod/tmp/socket/aa6712c41c455cd6c136c0af15d2345956ab9425e8690e4bd36845a59c384c83/atta: dial unixpacket /run/user/205013812/libpod/tmp/socket/aa6712c41c455cd6c136c0af15d2345956ab9425e8690e4bd36845a59c384c83/atta: connect: no such file or directory

[haircommander]

okay so I see now this was an unintended breaking change, and also easily reproducable. I will need to consult with the over developers on how best to handle it.

[haircommander]

thanks for your patience

[haircommander]

I've opened #226 to fix

[ksdd]

Great! Thank You.

[ksdd]

Hi @haircommander,

today I've got an update of conmon to version 2.0.24 via the mentioned repo. I've seen, that #226 was merged before 2.0.24 was released. So I hoped, this issue is fixed. Unfortunately this is not the case. I get the same error as before. Are there still fixes needed or was it supposed to be fixed in 2.0.24?

[haircommander]

unfortunately, the only way to preserve backward compatibility is to gate the new behaviour (that would fix your bug) behind a flag, meaning we also need a corresponding podman PR to fix your issue. that PR is here containers/podman#8979 and I'd like to make it in before podman 3.0 release :)

[ksdd]

Okay, thank you. So I will wait for a new podman release and stick to the older conman version meanwhile.