502 Bad Gateway after commit about docker-redmine HOT 18 CLOSED

@berndschneider
I would suggest you start the image with the bash command, i.e.

docker run -it [other-options] sameersbn/redmine:2.5.2-2 bash

make the required changes and then commit the image.

from docker-redmine.

For test purposes, I committed the image without making any changes, even then, I get "502 bad gateway", with this new created image...

from docker-redmine.

One addition. The unicorn-error log is telling me:

[2014-09-11T13:49:25.391000 #918] FATAL -- : error adding listener addr=/home/redmine/redmine/tmp/sockets/redmine.socket
/home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/socket_helper.rb:152:in `bind_listen': socket=/home/redmine/redmine/tmp/sockets/redmine.socket specified but it is not a socket! (ArgumentError)

    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/http_server.rb:242:in `listen'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/http_server.rb:809:in `block in bind_new_listeners!'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/http_server.rb:809:in `each'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/http_server.rb:809:in `bind_new_listeners!'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/lib/unicorn/http_server.rb:138:in `start'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/gems/unicorn-4.8.3/bin/unicorn_rails:209:in `<top (required)>'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/bin/unicorn_rails:23:in `load'
    from /home/redmine/redmine/vendor/bundle/ruby/2.1.0/bin/unicorn_rails:23:in `<main>'

from docker-redmine.

add to start script
rm -f /home/redmine/redmine/tmp/sockets/redmine.socket
or do it in container before commit it to new image

from docker-redmine.

or try to commit stopped container

from docker-redmine.

@berndschneider as pointed by @elderone when the image is commited the unicorn socket is also commit and for some reason this is causing an issue. That's why I had suggested to start the image with the bash command (which will not start unicorn and nginx) and then commit the image.

However if you have already overwritten the sameersbn/redmine image, then I would suggest you first pull the redmine image.

from docker-redmine.

@berndschneider Ideally this should not happen and it obviously needs some fixing. Above suggestions are workarounds.

from docker-redmine.

@sameersbn in new container from new image you can remove this regular file and restart it. there is no need to pull old image

from docker-redmine.

@berndschneider i just tried your use case and I dont see the issue.

docker run --name='redmine' -it [OPTIONS] sameersbn/redmine:latest
# [CTRL-C] to stop the container
docker commit redmine sameersbn/redmine
# now run the image again
docker run --name='redmine' -it [OPTIONS] sameersbn/redmine:latest
> Unlinking stale socket /var/run/supervisor.sock
> Installing plugins...
> Executing plugins startup script...
> Installing themes...
> Starting unicorn...
> Starting nginx...

It seems like you are probably seeing this issue because you are commiting a running container.

from docker-redmine.

@berndschneider Nope, committing a running container as well does not show that issue. Can you share the commands so that I can use to reproduce this issue?

from docker-redmine.

@elderone thats correct. I am suggesting pulling the image just to have a fresh start. Besides the pull is not really going to download anything. It will just retag the image.

from docker-redmine.

@berndschneider by the way, if I may ask what changes are you making to the image?

from docker-redmine.

Thanks for the fast replys. :-)

Every time I had this issue, I pulled the fresh image, and it was working again, until I tried to commit it.

I committed after stopping the container. The only difference I see is, that I started in detached mode (-d) and that I have mysql in a second container, the data path in the host system (as described in the readme file) and I have 2 themes in the theme directory, no plugins.

Later I will try your suggestions and tell you the results.

Thanks

from docker-redmine.

I had to change the Gemfile, because if I tried to install a plugin (aka copy it into the plugins directory), I got an error, that the version of installed rake (x.1.0) is different and lower then the one requested in the Gemfile (x.1.1).

To preserve a running system, I committed it, but unfortunately, the preserved system was not working anymore. :-)

from docker-redmine.

@berndschneider In that case, can you try deleting the tmp directory in the data volume and then start the image. This will probably resolve the issue you are having with the gem installation.

edit:
I have faced a similar issue before and simply deleting the tmp directory resolved the issue. But it might not work all the time. Can you tell me the plugin you are trying to install.

from docker-redmine.

@berndschneider try change Gemfile of plugin, but not of redmine
just comment out line with "bad" gem

from docker-redmine.

@berndschneider yep.. that is also a pluasible solution which has a very high probability of resolution, way higher than the deletion of tmp. 👍 @elderone

from docker-redmine.

Ok, here is a short status update.

1. Deleting redmine.socket using

rm -f /home/redmine/redmine/tmp/sockets/redmine.socket

before stopping the container does the trick. Now the new created image was working as expected.

2. The plugin I tried to install was "redmine_dmsf" - the installation was not working, because the gem xapian could not be installed (missing dependencies, like C++ compiler, ...). After commenting out this group in the dmsf-Gemfile, it was working.

So, in general, I have a solution. :-)

from docker-redmine.