-
Notifications
You must be signed in to change notification settings - Fork 264
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
alpine 3.13, armv7 network-access seems to be broken #135
Comments
I'm seeing similar issues:
I've checked on three Raspberry Pis and they're all presenting the same issue and showing the date (Similar discussion here: https://gitlab.alpinelinux.org/alpine/aports/-/issues/12091) |
Hi based on the post above ( https://gitlab.alpinelinux.org/alpine/aports/-/issues/12091) looks like another workaround would be to run the latest alpine:13.0 without any security profile (seccomp)
cheers |
Thanks, @jipp I'm running quite a lot of containers. Rather than upgrading and modify security policies, I downgraded everything to 3.12 and wait for a fix. |
Update: This was probably something local, it works fine again after reboot and on another machine. I've run into something that sounds similar to this when resolving $ docker run --rm -ti alpine:3.13.0
/ # ping codeload.github.com.
ping: bad address 'codeload.github.com.'
/ #
$ docker run --rm -ti alpine:3.12.3
/ # ping codeload.github.com.
PING codeload.github.com. (140.82.121.10): 56 data bytes
64 bytes from 140.82.121.10: seq=0 ttl=37 time=29.817 ms
64 bytes from 140.82.121.10: seq=1 ttl=37 time=30.078 ms
^C Querying 8.8.8.8 and my router (openwrt) seems to work but not using thru dockers DNS servers: $ docker run --rm -ti alpine:3.13.0
/ # apk add bind-tools
fetch https://dl-cdn.alpinelinux.org/alpine/v3.13/main/x86_64/APKINDEX.tar.gz
...
/ # host codeload.github.com 8.8.8.8
Using domain server:
Name: 8.8.8.8
Address: 8.8.8.8#53
Aliases:
codeload.github.com has address 140.82.121.10
/ # host codeload.github.com 192.168.1.1
Using domain server:
Name: 192.168.1.1
Address: 192.168.1.1#53
Aliases:
codeload.github.com has address 140.82.121.9
/ # host codeload.github.com
codeload.github.com has address 140.82.121.9
Host codeload.github.com not found: 3(NXDOMAIN)
Host codeload.github.com not found: 3(NXDOMAIN)
/ # cat /etc/resolv.conf
# This file is fetched from the host via vpnkit-bridge
nameserver 192.168.65.1 |
Found this to be an issue downstream and created an issue on a php image repo. tl;dr - it looks like the issue is ipv4/ipv6 related. Not sure if it's just curl or ipv6 being blocked across the entire container. You can currently make successful requests by using the --ipv4 flag |
I experienced this exat same issue when 3.12.0 was released, but only ir ARMv7 Seems the root cause is related to clock/time not being properly set: $ docker run -it alpine:3.13.0 Note that as time is not correct, cannot validate TLS/SSL, also seems not only wrong, indeed seems is not even "ticking" |
Looks like the issue for me was that I needed to update to Docker for Mac 3.1.0. |
You need to update libseccomp on your host to 2.4.2 or newer and Docker to 19.03.9 or newer (see moby/moby#40734). |
Following the wiki I got it running. Nevertheless this would mean that running the default Raspbian 32bit (armv7) this is a mandatory step for Alpine images to be used. |
@TBK thanks. I'm running dietpi which is still based on stretch. Afaik there is no way to install seccomp with that version. Feels like to get this straight without any hacky-solution a distro-change seems mandatory. 😞 |
As this is not a bug, but a known problem with old versions and possible workarounds I will close this issue. Thanks for your help! |
Why is this closed ?! |
This is problem of the docker host library versions.
|
Hello, Today i format one of two rasp (raspberry pi 1 B+ armv6l) because i had strange kernel panic preventing reboot. I setup again raspbian buster 10, setup docker and container start but no resolution again. ping into container give error
and
So my question do you think something had changed related to libseccomp2 and docker dns resolution problem so that workaround don't work anymore ? Do you think my hypothesis about apt crash related to apt version 1.8.2.3 (armhf), available to give you additional information Thanks in advance. edit : i found this apparently am not alone with kernel panic issue https://gitlab.alpinelinux.org/alpine/aports/-/issues/12091#note_147886 :) edit 2 : |
I did change the base image to Since I run the build on cloud pipelines, such as github action, travis ci, I can't do anything to the build agents. What can I do to fix this issue from my end? ====
|
@ozbillwang I had the same problems. I pinned the last working version and started to migrate my images to debian. |
ic. So the problem is only in Ubuntu. Thanks for the update. |
@ozbillwang Sorry for being unclear. I had the same problems on different host systems (Debian Arm, Ubuntu x86, CI-providers). I am pretty sure you can update your Ubuntu to a libseccomp version that is compatible, so it’s no Ubuntu-specific problem. I derive my own built docker images from now on on debian instead of alpine. |
Do you think this is related to inner container alpine os ? because this is host libseccomp version and host apt which stop working in my case. Maybe other inner container os don't use libseccomp ? |
ok, so you change the Thanks for the clarification |
Hey!
Some of my nightly armv7 alpine builds suddenly started failing. It looks like there is a problem on installing packages through apk on alpine 3.13 on armv7l.
Unfortunately this means that the latest tag is currently broken for armv7l.
Example commands and output:
Same thing is working perfectly fine on alpine 3.12:
I started digging deeper. It looks like the network access on the docker container seems to be broken.
When running the container privileged the problems are gone:
The text was updated successfully, but these errors were encountered: