If anyone is able to run Discourse using docker in a M1 Mac, please let me know. In the mean time, I’ll try to follow the other guide! Thank you!
I tried briefly today, and got stuck at the same step as you, but with an error:
Invalid gemspec in [/usr/local/lib/ruby/gems/2.7.0/specifications/default/zlib-1.1.0.gemspec]: Malformed version number string specification_version
bundler: failed to load command: rake (/usr/local/bin/rake)
Yes, please do. There are multiple team members using Discourse on M1 (myself included) every day, so it works quite well!
Let us know if you run into any issues.
Thank you very much for your time and help! At least I’m not the only one stuck with it.
Hi, i think we should make a description about the Ember-CLI here, and a shourcut for below command without enter into docker container.
And i can’t get it work by execute above commands into container, seems container didn’t expose the 4200 port.
Manual expose the port 4200 by edit d/boot_dev
:
After reboot container I access localhost:4200
got this:
discourse@discourse:/tmp$ cat error.dump.cab4cc444229d44fc0fce2deb8695880.log
=================================================================================
ENV Summary:
TIME: Sun May 02 2021 08:01:12 GMT+0000 (Coordinated Universal Time)
TITLE: ember
ARGV:
- /usr/bin/node
- /src/app/assets/javascripts/node_modules/.bin/ember
- server
- --proxy
- http://localhost:3000
EXEC_PATH: /usr/bin/node
TMPDIR: /tmp
SHELL: /bin/bash
PATH:
- /tmp/yarn--1619942449179-0.1910991449592403
- /src/app/assets/javascripts/discourse/node_modules/.bin
- /home/discourse/.config/yarn/link/node_modules/.bin
- /src/app/assets/javascripts/node_modules/.bin
- /home/discourse/.yarn/bin
- /usr/libexec/lib/node_modules/npm/bin/node-gyp-bin
- /usr/lib/node_modules/npm/bin/node-gyp-bin
- /usr/bin/node_modules/npm/bin/node-gyp-bin
- /usr/local/sbin
- /usr/local/bin
- /usr/sbin
- /usr/bin
- /sbin
- /bin
PLATFORM: linux x64
FREEMEM: 8603062272
TOTALMEM: 41962496000
UPTIME: 612639
LOADAVG: 3.32177734375,2.19580078125,1.47900390625
CPUS:
- Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz - 3301
- Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz - 3300
- Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz - 3300
- Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz - 3301
ENDIANNESS: LE
VERSIONS:
- ares: 1.15.0
- brotli: 1.0.7
- cldr: 35.1
- http_parser: 2.9.3
- icu: 64.2
- modules: 64
- napi: 7
- nghttp2: 1.41.0
- node: 10.23.0
- openssl: 1.1.1g
- tz: 2019c
- unicode: 12.1
- uv: 1.34.2
- v8: 6.8.275.32-node.59
- zlib: 1.2.11
ERROR Summary:
- broccoliBuilderErrorStack: [undefined]
- code: ECONNREFUSED
- codeFrame: [undefined]
- errorMessage: connect ECONNREFUSED 127.0.0.1:3000
- errorType: [undefined]
- location:
- column: [undefined]
- file: [undefined]
- line: [undefined]
- message: connect ECONNREFUSED 127.0.0.1:3000
- name: Error
- nodeAnnotation: [undefined]
- nodeName: [undefined]
- originalErrorMessage: [undefined]
- stack: Error: connect ECONNREFUSED 127.0.0.1:3000
at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1107:14)
=================================================================================
After edit bin/ember-cli
PORT to 3000
to 9292
all working.
Seems the ember-cli
bash unable work with ENV["UNICORN_PORT"]
.
The Ember CLI is a new (and hard fought) development; @eviltrout should be able to comment on that soon.
Yes this will have to be updated. In the meantime you can disable that by setting the NO_EMBER_CLI
environment variable to 1
.
Probably obvious but could you clarify where you set the env variable @eviltrout ?
I tried in the d/unicorn file like this:
docker exec -it -u discourse:discourse discourse_dev /bin/bash -c "$CMD" -e NO_EMBER_CLI=1
…but that didn’t work (still getting " Ember CLI is Required in Development Mode" message at localhost:9292).
d/boot_dev -e NO_EMBER_CLI=1
I tried this today, and also ran into issues. The error I saw was because Docker’s architecture emulation doesn’t support inotify
(which we use a lot in Discourse development). For now, I’ve added a warning to d/boot_dev
when a non-x86_64 architecture is detected:
❯ d/boot_dev
WARNING: Docker architecture is not x86_64.
Discourse development is unlikely to work using Docker's architecture emulation.
Please try a native development installation.
I’ve now added a d/ember-cli
helper, and forwarded port 4200 by default. The info at the top of this topis has also been updated. Once you’ve updated, run d/rails s
in one terminal, and d/ember-cli
in another. I’ve also set NO_EMBER_CLI
as one of the variables which is passed through to Docker, so that is available if needed.
@david probs inconsequential but just FYI: The boot_dev script prints out a false error finding on the x86_64 check when I accidentally ran without Docker on…(but the ‘Is the docker daemon running’ part is right)…
WARNING: Docker architecture is not x86_64.
Discourse development is unlikely to work using Docker's architecture emulation.
Please try a native development installation.
...(snip)...
Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?
Thanks for this image, and the super-clear instructions!
I got psql: error: FATAL: Peer authentication failed for user "postgres"
when running d/boot_dev --init
.
Although the pg_hba.conf
in data/postgres/
had all the authentication methods set to “trust”, there was another one in /etc/postgresql/13/main/
which had the defaults (“peer” / “md5”).
I edited /etc/postgresql/13/main/pg_hba.conf
, changed all the methods to trust
, did d/shell
and ran sv restart postgres
to pick up the changes – and was able to continue by running d/bundle install; d/rake db:migrate RAILS_ENV=development; d/rake admin:create
manually.
Leaving here in case it’s helpful for anyone else!
Thank you for the guide.
However I’m having issues with creating a backup from the admin section.
The error I’m getting is:
pg_dump: error: connection to database "discourse development" failed: FATAL: Peer authentication failed for user "postgres".
I checked the pg_hba.conf file, and have all options set to trust.
Would be great i could get some assistance on how to get this to work.
I tried on Ubuntu as well as on MacOSX. Everything is working fine on both (creating Posts, API…) except for the backup functionality.
Thanks for this fantastic docker solution.
For running plugin specs, this works nicely:
d/rake plugin:spec["discourse-follow"]
Is there a way to target specific plugin tests, as per a non-docker dev?, e.g.:
LOAD_PLUGINS=1 RAILS_ENV=test rspec plugins/discourse-follow/spec/requests/follow_controller_spec.rb:32
I’ve tried e.g.:
d/rspec plugins/discourse-follow/spec/lib/updater_spec.rb:10 LOAD_PLUGINS=1 RAILS_ENV=test
But getting an error.
LOAD_PLUGINS
and RAILS_ENV
must precede the command to assign environment variables. Following the command, they’re being treated as arguments to rspec
, which it doesn’t understand.
LOAD_PLUGINS=1 RAILS_ENV=test d/rspec plugins/discourse-follow/spec/lib/updater_spec.rb:10
No, I’m not convinced that works - have you actually tried it?
I get:
NameError:
uninitialized constant Follow
I suspect this is because the environment variables are not being passed into the docker process?
That was why I was making them arguments.
Apologies, I misinterpreted your two commands as “the first works for this other spec, the second doesn’t work for this spec”. I don’t have a development environment set up to test.
Looking at the rspec file in Github, I think you’re right that it’s not passing in the environment variables. It looks like you should be able to run d/shell
to get a shell inside the container, then run your rspec command there though.
Simon, that’s more than an excellent useable workaround, thanks!
Just tried that and it works, ie:
my-blah-machine:~/discourse$ d/shell
discourse@discourse:/src$ LOAD_PLUGINS=1 RAILS_ENV=test rspec plugins/discourse-follow/spec/lib/updater_spec.rb:37
Have added this and the whole plugin version to the Wiki
Taking a closer look at d/exec
, which both d/shell
and d/rspec
use, I think it can also be shortened like so:
RAILS_ENV=test d/exec LOAD_PLUGINS=1 rspec plugins/discourse-follow/spec/lib/updater_spec.rb:37
d/exec
does pass in RAILS_ENV
but not LOAD_PLUGINS
, hence them being on either side of d/exec
.
That gives me an error:
OCI runtime exec failed: exec failed: container_linux.go:380: starting container process caused: exec: "LOAD_PLUGINS=1": executable file not found in $PATH: unknown
Ah, I guess environment variables can’t be used with docker exec
like that. Oh well, at least opening the shell first works.