20 November 2016 at 01:57, Luís Picciochi Oliveira <pitxy...@gmail.com> wrote: > While using Docker, some commands on some images stopped working. For example: > > $ docker run -it --rm centos:centos6 bash > This does not work. It always returns with code 139: > > $ echo $? > 139 > > > ...but the following does work! > > $ docker run -it --rm centos:7 bash > [root@06c8e740c5bc /]# exit > $ echo $? > 0 > > It's not only the bash command. For example: 'chmod', 'sh' and a few others > also fail with the same error code. 'echo', 'ls', 'cat' and others work > without > any issues. > > I found out that using the previous kernel (linux-image-4.7.0-1-amd64 > (4.7.8-1)), I could run all commands on all images.
This is very bizarre, and sounds similar to https://github.com/tianon/docker-brew-debian/issues/55 which was filed recently as well. I wonder if the kernel configuration for 4.8 in Debian has changed something related to backwards compatibility? As you can see on that issue, a colleague of mine is running a 4.8 kernel on Gentoo and can't reproduce, but I can on my Debian Unstable system. :( > The only report I found online about this was the following: > https://forums.docker.com/t/docker-run-it-has-started-failing-with- > status-139/18309 > > Coincidentally, it just had a reply yesterday, which seems to match exactly my > experience. Definitely interesting, although sadly lacking in debugging tips, and I'm afraid I don't have any great ideas to help debug this either. :( Perhaps it would be helpful to try reproducing with a wheezy chroot (taking Docker out of the equation entirely)? ♥, - Tianon 4096R / B42F 6819 007F 00F8 8E36 4FD4 036A 9C25 BF35 7DD4
