Auplink - Signal: segmentation fault (core dumped)

I am trying to get an app that relies on the directory on the host: /lib/modules/4.14.39/ and the device on the host: /dev/i2c-1, which are both available on the host. I am using the latest resinOS and also attempted to use a python resin image, since the app is python-based. I am working with a RPi3.

I can see the following message and then the container goes into a restart loop:

Jul 31 04:28:47 04fd668 healthdog[762]: time="2018-07-31T04:28:47.279490698Z" level=warning 
msg="Couldn't run auplink before unmount 
/var/lib/docker/aufs/mnt/d7af73be3ac247993662b692d5c5ed1730e4e08286d485ce9bfdbfc702ac440e: 
signal: segmentation fault (core dumped)"

...
31.07.18 00:37:53 (-0400) Service exited 'main 
sha256:147eacd58ddc9c18994e20ef05b9f038e1ac875efb19266e7c804bc2930e1

This is what I see on the host:

root@04fd668:~# ls -la 
/var/lib/docker/aufs/mnt/d7af73be3ac247993662b692d5c5ed1730e4e08286d485ce9bfdbfc702ac440e*
/var/lib/docker/aufs/mnt/d7af73be3ac247993662b692d5c5ed1730e4e08286d485ce9bfdbfc702ac440e:
total 8
drwxr-xr-x  2 root root 4096 Jul 31 04:27 .
drwx------ 36 root root 4096 Jul 31 04:27 ..

/var/lib/docker/aufs/mnt/d7af73be3ac247993662b692d5c5ed1730e4e08286d485ce9bfdbfc702ac440e-init:
total 8
drwxr-xr-x  2 root root 4096 Jul 31 04:27 .
drwx------ 36 root root 4096 Jul 31 04:27 ..

It seems like the host is trying to run “auplink” but it is providing a segmentation fault:

root@04fd668:~# auplink 
Segmentation fault (core dumped)

I can’t seem to get around this. Any ideas?

We reported this as bug earlier in the month: https://github.com/resin-os/resin-qemu/issues/88

Supposed to be fixed in the next release, perhaps Resin people can comment on the timeframe.

1 Like

Hi @nmajin @splitice,

This issue should be addressed by the move from aufs to overlayfs. The fix has already been merged (https://github.com/resin-os/resin-qemu/blob/master/CHANGELOG.md#v2140rev1) but we haven’t released the image yet - we will update this thread as soon as we do.

Thanks,
Kostas

1 Like

Is there any workaround? I am looking to provide a demo of an app via resin’s CICD pipeline and was really hoping to showcase this.

I too would like to see a development image pushed. Preferably with our Squashfs support PR merged as well.

Pretty much for the same reason as nmajin since our CICD pipeline is based around using QEMU and emulated hardware devices to automate our application testing. This is currently held up behind regular crashes of Auplink and lack of kernel Squashfs support.

Hi,

Sorry for the late response here, squashFS is now supported for QEMU balenaOS versions >= 2.14.1+rev1

Thanks,
Kostas