Docker: cannot execute oxidized

[72tlukas72]

Hello,
I tried upgrade Oxidized from 0.30.1 to 0.31.0, but after start I got only error mesage cannot execute oxidized

Log:
*** Running /etc/my_init.d/00_regen_ssh_host_keys.sh...
*** Running /etc/my_init.d/10_syslog-ng.init...
Dec 2 11:47:51 f84c882441e4 syslog-ng[14]: syslog-ng starting up; version='4.3.1'
*** Booting runit daemon...
*** Runit started as PID 24
Dec 2 11:47:52 f84c882441e4 cron[30]: (CRON) INFO (pidfile fd = 3)
Dec 2 11:47:52 f84c882441e4 cron[30]: (CRON) INFO (Running @reboot jobs)
setuser: cannot execute oxidized: [Errno 2] No such file or directory
setuser: cannot execute oxidized: [Errno 2] No such file or directory
setuser: cannot execute oxidized: [Errno 2] No such file or directory
setuser: cannot execute oxidized: [Errno 2] No such file or directory

[robertcheramy]

Can you provide more information on your setup (how do you start the container, with which arguments...)?
I use podman and so it will help me to setup docker correctly so that I can reproduce your problem ;-)

[72tlukas72]

I'm using Portainer, nothing change (beside version if image). Command is '/sbin/my_init' and mapped directory for config, data and model.

[robertcheramy]

I see several problems in the Dockerfile:

• bd8d579 introduced the creation of /home/oxidized/.config/oxidized/, owned by root
• f02d689 stated in README.md the option --user oxidized which does not work without extra setup
• there is an ubuntu user in the image, which should not be there
  I will look after this.

I can run the docker image with docker without problem:

$ sudo docker run -i --rm -v /tmp/oxidized:/home/oxidized/.config/oxidized -p 8888:8888/tcp -t oxidized/oxidized:latest
*** Running /etc/my_init.d/00_regen_ssh_host_keys.sh...
*** Running /etc/my_init.d/10_syslog-ng.init...
Dec  5 06:45:45 9b83584d0ed1 syslog-ng[14]: syslog-ng starting up; version='4.3.1'
*** Booting runit daemon...
*** Runit started as PID 24
Dec  5 06:45:46 9b83584d0ed1 cron[31]: (CRON) INFO (pidfile fd = 3)
Dec  5 06:45:46 9b83584d0ed1 cron[31]: (CRON) INFO (Running @reboot jobs)
I, [2024-12-05T06:45:46.926363 #30]  INFO -- : Oxidized starting, running as pid 30
I, [2024-12-05T06:45:46.926948 #30]  INFO -- : lib/oxidized/nodes.rb: Loading nodes
I, [2024-12-05T06:45:46.992011 #30]  INFO -- : lib/oxidized/nodes.rb: Loaded 1 nodes
I, [2024-12-05T06:45:47.993645 #30]  INFO -- : Configuration updated for /fw4

Also using /sbin/my_init as an entry point works:

$ sudo docker run -i --rm -v /tmp/oxidized:/home/oxidized/.config/oxidized -p 8888:8888/tcp -t oxidized/oxidized:latest /sbin/my_init
*** Running /etc/my_init.d/00_regen_ssh_host_keys.sh...
*** Running /etc/my_init.d/10_syslog-ng.init...
Dec  5 06:48:22 58af4f62aa31 syslog-ng[14]: syslog-ng starting up; version='4.3.1'
*** Booting runit daemon...
*** Runit started as PID 24
Dec  5 06:48:23 58af4f62aa31 cron[30]: (CRON) INFO (pidfile fd = 3)
Dec  5 06:48:23 58af4f62aa31 cron[30]: (CRON) INFO (Running @reboot jobs)
I, [2024-12-05T06:48:23.757523 #31]  INFO -- : Oxidized starting, running as pid 31
I, [2024-12-05T06:48:23.758047 #31]  INFO -- : lib/oxidized/nodes.rb: Loading nodes
I, [2024-12-05T06:48:23.818257 #31]  INFO -- : lib/oxidized/nodes.rb: Loaded 1 nodes
I, [2024-12-05T06:48:24.820495 #31]  INFO -- : Configuration updated for /fw4

To further inspect your problem, I really need the arguments you are passing to docker, in order to try to reproduce.

[72tlukas72]

You can study now... What should be change for upgrade from 0.30.1 to 0.31? In release notes is nothing for change.

[robertcheramy]

I've found the difference. In 0.31.0, oxidized gets installed in /usr/local/bundle/bin/ instead of /usr/local/bin/.

The path is correctly specified in the container image, which explains why it works for me:

$ podman inspect docker.io/oxidized/oxidized:0.30.1 | grep PATH
                    "PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
$ podman inspect docker.io/oxidized/oxidized:0.31.0 | grep PATH
                    "PATH=/usr/local/bundle/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
                    "created_by": "ENV PATH=/usr/local/bundle/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",

Portainer probably did not updated its specified path when you tried the new image, which causes the error on your server.

A quick workaround would be to update the PATH in Portainer.

[robertcheramy]

The change was introduced in commit c56bc4b (PR #3272).

I tend to leave this unchanged: introducing /usr/local/bundle as GEM_HOME helps reduce the container size by installing the gems in a prebuilder-container and copying /usr/local/bundle into the oxidized container.

Portainer should be able to re-read the PATH from the containter image, or let you set the PATH correctly. Could you try this?

[E-t-z]

I tried to update path, but Container still fails to start, now it logs:

2024/12/05 22:52:08,stdout,/usr/lib/ruby/vendor_ruby/rubygems.rb:264:in `find_spec_for_exe': �[1mcan't find gem oxidized (>= 0.a) with executable oxidized (�[1;4mGem::GemNotFoundException�[m�[1m)�[m
2024/12/05 22:52:06,stdout,	from /usr/local/bundle/bin/oxidized:25:in `<main>'
2024/12/05 22:52:06,stdout,	from /usr/lib/ruby/vendor_ruby/rubygems.rb:283:in `activate_bin_path'
2024/12/05 22:52:06,stdout,/usr/lib/ruby/vendor_ruby/rubygems.rb:264:in `find_spec_for_exe': �[1mcan't find gem oxidized (>= 0.a) with executable oxidized (�[1;4mGem::GemNotFoundException�[m�[1m)�[m
2024/12/05 22:52:06,stdout,	from /usr/local/bundle/bin/oxidized:25:in `<main>'

[E-t-z]

Running it on Synology if it matters and using the latest image.

[E-t-z]

Nevermind, creating Container from scratch updated PATH and added GEM_HOME (which was missing from my config)
It would be great if Documentation would reflect such changes.

[72tlukas72]

Same problem in v0.32.0

[robertcheramy]

And probably same solution - re-read your container.

[72tlukas72]

What is "re-read" meaning?

[robertcheramy]

The Docker container is OK, but with the update it introduces an environment Variable GEM_HOME, wich is set in the container. The problem on this issue was, that the container running engine did not notice the new environment, which causes the problem.

If this is not your problem, create a new issue, and be sure to read https://github.com/ytti/oxidized/blob/master/docs/Issues.md first.

[72tlukas72]

So what path should be set in GEM_HOME? In docs is nothing about this variable.

[rmcmilli]

Came looking for info about the ubuntu user. Looks like ubuntu made a decision to include the ubuntu user with their base image and the only way around it is to either remove the user on build or use the ubuntu user instead of a custom one. https://askubuntu.com/questions/1513927/ubuntu-24-04-docker-images-now-includes-user-ubuntu-with-uid-gid-1000

Edit: Looks like the phusion image used gives a disclaimer: https://github.com/phusion/baseimage-docker/releases/tag/noble-1.0.0

[robertcheramy]

@72tlukas72 - please read the whole Issue, everything is explained in it. I'll produce a FAQ some time, and this is why I assigned the issue to me.
@rmcmilli - Thanks for the information. I've been thinking about migrating to debian-slim [or provide two images], but the effort involved stood so far in the way ;-)

[72tlukas72]

Okay I will wait to v0.33 :-D

[robertcheramy]

I've reverted the change in PR #3429, which will fix this issue.