frntn / docker-cloud9

:whale: Web-based IDE inside a docker container

frntn/docker-cloud9

Web-based Cloud9 IDE container

docker-cloud9's Issues

user namespace not working as described in blog

Thanks for the blog. This inspired me to learn docker. So... I am docker noob

I started with these steps and wanted to build this myself since.. you know I am new and want to learn docker.

I am having trouble when I docker build when the docker daemon is running with a valid user

% cat /etc/subuid
dragon:1000:1000
dockremap:165536:65536

% cat /etc/subgid
dragon:1000:1000
dockremap:165536:65536

and when I have the daemon as

sudo /usr/bin/docker daemon --userns-remap=dragon

and execute

% ls -latr                                                                                                                                                                                                               16-03-08 - 10:51:09
total 24
drwxrwxrwx 4 dragon dragon 4096 Mar  7 09:31 ..
-rwxrwxrwx 1 dragon dragon  204 Mar  7 19:57 supervisord.conf
-rwxrwxrwx 1 dragon dragon  691 Mar  8 05:44 Dockerfile-mod
-rwxrwxrwx 1 dragon dragon 1053 Mar  8 05:48 Dockerfile
-rwxrwxrwx 1 dragon dragon 2709 Mar  8 10:23 .alias
drwxrwxrwx 2 dragon dragon 4096 Mar  8 10:51 .


% docker build -t rrag/cloud9 . 

after a few mins of lots of things getting printed I see

:Compiling ncurses...
tar: ncurses-5.9/test/inchs.c: Cannot change ownership to uid 1001, gid 0: Invalid argument
tar: ncurses-5.9/test/ncurses_tst.hin: Cannot change ownership to uid 1001, gid 0: Invalid argument
tar: ncurses-5.9/test/keynames.c: Cannot change ownership to uid 1001, gid 0: Invalid argument
....

and the build aborts.

Now when I stop and restart the daemon with

sudo /usr/bin/docker daemon --userns-remap=default

The whole build is successful. and I am able to launch c9, and the .c9 is created with 165536 as expected.

It got me thinking, may be the build wont work, so let me pull down your image

again

sudo /usr/bin/docker daemon --userns-remap=dragon

and

% docker pull frntn/cloud9                                                                                                                                                                                               16-03-08 - 10:57:43
Using default tag: latest
latest: Pulling from frntn/cloud9
7268d8f794c4: Pull complete 
a3ed95caeb02: Pull complete 
d9a49bc2b1b0: Pull complete 
b965864d2d45: Pull complete 
47bed597ecf4: Pull complete 
432750a489d5: Pull complete 
ad86930c50cb: Pull complete 
c7743c99b67e: Pull complete 
caabee7d9368: Extracting [==================================================>] 102.1 MB/102.1 MB
22ec05f47241: Download complete 
771be47a7e83: Download complete 
8a6ffb3d8644: Download complete 
4d0c718f85a2: Download complete 
6e618d1b0094: Download complete 
8ca9988a6281: Download complete 
037a649b005a: Download complete 
b35450747bf8: Download complete 
f7d5cf41f0cc: Download complete 
failed to register layer: Untar re-exec error: exit status 1: output: Container ID 1001 cannot be mapped to a host ID

well, may be there was a problem with the way you pushed it up. let me push my own and pull that and see what happens.

again same result.

I was out of ideas, reading all the documentation leads me to believe your approach is correct. but still no luck.

Here is how I hacked it to work

% pwd
/var/lib/docker

% ls -latr
total 20
drwxr-xr-x 64 root   root   4096 Mar  6 13:01 ..
drwx------  9 dragon dragon 4096 Mar  7 10:24 1000.1000
drw-------  3 root   root   4096 Mar  8 10:30 network
drwx-----x  5 root   root   4096 Mar  8 10:35 .
drwx------  9 165536 165536 4096 Mar  8 10:35 165536.165536

now

% sudo su
% rm -rf 1000.1000
% mv 165536.165536 1000.1000
% chown dragon:dragon 1000.1000
% cd 1000.1000

change the ownership to dragon:dragon for those which were owned by 165536.165536

Now restart docker daemon with

sudo /usr/bin/docker daemon --userns-remap=dragon

docker images

I see everything !!!, and it works when I start c9 and the .c9 folder is created with dragon:dragon

I realize what I have done is basically rename 165536 as 1000 and updated the right permissions, and this is certainly a hack.

I would love to hear from you if there is a better solution or may be you missed to add a step in your blog. I keep coming across newuidmap when ever I search for /etc/subuid so is it likely you missed to update a step in your blog?

Thanks again for this article, it is well written and got me interested to learn docker.