Docker snap 18.09.9 error

Hi all,

It looks like 18.09.9 was promoted to the stable channel yesterday. Unfortunately, it’s broken on my system and I’m not sure where else to report the bug.

snap info shows the contact as snappy-devel@lists.ubuntu.com but that list seems to no longer exist.

$ snap version
snap    2.42.1
snapd   2.42.1
series  16
ubuntu  19.10
kernel  5.3.0-23-generic

After the refresh, my syslog was spammed with the following error, repeatedly, until it filled my disk and crashed my machine:

Nov 15 07:51:19 ronin docker.dockerd[8055]: time="2019-11-15T07:51:19.032482136-08:00" level=error msg="failed to get event" error="rpc error: code = Unavailable desc = all SubConns are in TransientFailure, latest connection error: connection error: desc = \"transport: Error while dialing dial unix /run/containerd/containerd.sock: connect: permission denied\"" module=libcontainerd namespace=plugins.moby     

The offending directory:

$ sudo ls -la /run/containerd
total 0
drwx--x--x  5 root root  120 Nov 14 22:09 .
drwxr-xr-x 53 root root 1700 Nov 15 07:54 ..
srw-rw----  1 root root    0 Nov 14 22:08 containerd.sock
drwx--x--x  2 root root   40 Nov 14 22:08 io.containerd.runtime.v1.linux
drwx--x--x  2 root root   40 Nov 14 22:08 io.containerd.runtime.v2.task
drwx------  3 root root   60 Nov 14 22:09 runc

Removing and reinstalling the snap produces the same errors.

1 Like

@tianon can you take a look at this?

… and also fix the contact field in the snap to at least point to something existing (like this forum) …

Yeah, I’ll take a look – this sounds like it’s closely related to / the same issue as what’s been reported over in https://bugs.launchpad.net/bugs/1852720 although there’s maybe more there to look into given the claim the Docker snap was auto-installed?

I’ve managed to reproduce – it’s really containerd installed on the host which is causing the underlying issue. When containerd runs on the host, it creates /run/containerd/containerd.sock, which dockerd inside Snappy can see and thus thinks it needs to use, but Snappy (rightfully) blocks it from doing so. Working on a fix now.

It may only be coincidence, but the snap refreshed yesterday, shortly before the errors began (and filling 30G+ of syslog before running out of disk space). The snap previously had been installed for months without issue.

Thanks for the help tracking it down!

1 Like

I’ve published my fix to the candidate channel if you wanted to give it some testing before I go to stable. :+1:

Installed and confirmed that the error is resolved. Thanks!

Thanks for testing and confirming! Fix is now released to the stable channel.

1 Like