I'm using IntelliJ Idea with the Rubymine plugin, and Heroku plugin, and I'm a bit stumped when it comes to pushing an app to Heroku. I can set it up and actually push the app: off it goes, and launches successfully. But even before I did that I was expecting it to fail because I couldn't find anyway of getting it to do a db:migrate. And so it proved: the app is launched but of course it fails pretty much immediately because there's no DB.
I've looked on the forums, google etc but nothing. Is this something that can only be done from the command line? I would have expected an IDE to be what it says on the tin: Integrated. So I feel I'm missing something. The Run configuration works as I've said, and the Heroku log has no errors, but I need a way to tell it to run the db:migrate before launching.
Any ideas?
The answer is that it is something that it doesn't do - confirmed by IntelliJ. So in this case, one needs to push the app to Heroku, drop to the command line and run the migration there: heroku run rake db:migrate.
I raised a feature request so this may be something that can be added in the future.
I haven't figured out how to do this automatically as part of the RubyMine/JetBrains Heroku Plugin. But you can cobble this together from other sources.
Start with this gist that creates rake tasks for Heroku operations provided by this answer to a similar question:
Then create a new run configurations for each Rake task that you will be using at some point. You will need at minimum push and migrate. But the other tasks might interest you.
Then create a new compound run configuration calling the Rake tasks you created in step 2 in the correct order.
Step 2 and 3 can be consolidated by creating one rake task for migration, and adding a before rake task to push. But that's kind of counter intuitive.
Note: that this approach does require you to have the Heroku CLI installed and configured with valid credentials.
Related
I didn't use Heroku for a while. I found Heroku change something so want to give it another try.
But after I click the "Deploy Branch" button, my app is still not working.
So I check the build log and realize Heroku seems not do the db:migrate command.
But it did do the asset:compile command. And I don't found anywhere to click to do the db:migrate thing.
So I have to do it with command line tools, right?
This is a well known limitation of Heroku. It won't run your migrations out of the box. However, you can automate it in couple of ways:
You can write a simple script that will first push new code to Heroku git repository and then run migrations. The problem is that you need to run this script locally on your machine
You can add this buildpack and then set environmental variable DEPLOY_TASKS to db:migrate. You can do this via UI, command line heroku config:set DEPLOY_TASKS='db:migrate' or you can add everything to app.json so it should work out of the box with deploy button.
You can use release phase by adding release: rake db:migrate to your Procfile.
Please keep in mind that there are many issues related to migrating your database during deployment. You can read about it in the docs for release phase.
I'm reading railstutorial.org Chapter 2 (about toy_app) using cloud IDE and I've run into a problem. I'm newbie so I would appreciate a simple sollution and explanation what I did wrong.
I followed every step as described in tutorial: I created my 'hello world' index page and I pushed it to heroku. I renamed my heroku site as well (it wasn't required in tutorial). Everything was fine at that moment.
I added user and micropost models. I've started a local server (on second terminal by typing rails server -b $IP -p $PORT) and added some users and microposts. Again, everything was working as expected.
I didn't close local server. I committed changes to git and I pushed it to Bitbucket and then to heroku. Finally, I typed heroku run rake db:migrate. It was the final step of this tutorial.
Everything seemed to be Ok, but I noticed that posts and users I added on local server weren't visible on my heroku site. Since I didn't close my local server, I checked it and posts I created were still available. As I said, situation on my heroku site was different.
I didn't get any errors during that process, only one warning after typing git push: Warning: Permanently added the RSA host key for IP address ... to the list of known hosts. Next time it didn't show up.
After my first try of every step I got promising outputs. After second and next tries it returned
On branch master
Your branch is up-to-date with 'origin/master'.
nothing to commit, working directory clean`
and
Everything up-to-date
So still nothing worrying.
I tried to restart server, to follow those steps again after closing my local server but it didn't help. I searched for solution but only what I found were problems with massive errors or different problems.
Actually, It is problem I can go along with because I don't need it updated at this stage, but I would like to hear an answer and find out if there is something wrong with heroku or it it normal behavior or even if it is newbie's blindness.
First your local database and heroku database are different. Any user created on local app won't be available on heroku app unless you use a common database.
rake db:migrate, in its most basic form, just runs the change or up method for all the migrations that have not yet been run. If there are no such migrations, it exits. It will run these migrations in order based on the date of the migration.
Try creating user and post on heroku app, it should work all fine.
I recently set up the Scheduler add on and set up my rake task, 'rake cron_jobs:my_task'.
When I test it with
'heroku run rake cron_jobs:my_task', it works fine.
The scheduler also claims it ran when it was supposed to, and is scheduled to run again, but there's no logging associated with the process the way https://devcenter.heroku.com/articles/scheduler#inspecting-output says there should be.
'heroku ps' shows no scheduled dynos, 'heroku logs --ps scheduler.1' has no output.
What am I missing?
Actually I was trying to solve this myself, and did not find the answer anywhere, so here it is if someone else is struggling with this:
heroku logs --tail --ps scheduler
--tail is important to keep streaming the logs.
My best guess: the heroku ps and heroku logs commands only give you status logs for currently running processes/dynos.
So after the scheduled rake task is done, you can't reach the logs through the command line.
You can access the history of your logs by using one of the logging addons. Most of them offer a free tier too.
They all are based on the log drains which you also could use directly, if you want to build it yourself.
Here is what I do for that:
Simply in your tasks itself include put statements to know when the job started running and when it is finished as well.
Also, you can include puts statement in the executed job as well.
I'm using paper trial add-on which is a very powerful logging tool that you can search and find any particular log at a specific time. Also, you can add an alert when your schedule job started to run.
I had a similar problem (using the newer Heroku PGBackups Service) and found an unexpected explanation for it.
The rake task rake pgbackups-archive was not run by Heroku Scheduler, but it worked when I ran it manually from the command line.
In my case, I noticed that my issues were caused by the different time zone used by the Heroku interface (which seems not to be CET). So my rake task which should have run at a specific time daily effectively never ran, as I changed the specific time throughout the day for testing and I always missed the specified time in the Heroku timezone.
You can try running the task every ten minutes and see if it works.
I have been trying to deploy a rails application using capistrano. However I am not able to reach the end of deployment and there is no way to figure out why. The script returns an error code 256 and stops at the following line in the deploy script
Command bundle exec rake assets:clean && EXECJS_RUNTIME='Node' JRUBY_OPTS='-J-d32 -X-C' bundle exec rake assets:precompile returned status code 256
There is no more explanation to it. Has anybody faced the similar issue while deploying on windows?
You should provide more information on the context of your situation. For instance, show the relevant code from your deploy.db file. That could tell us more about a possible problem source.
As a tip, when facing capistano deploying bugs, use the debugging flag to get more verbose output, and step through the process.
cap deploy:cold -d
Anywho... I was facing a similar problem, while optimizing my capistrano deploying times. In case you only want to modify your assets:precompile task, to include your EXECJS_RUNTIME and JRUBY_OPTS custom values, you could try doing this.
set :jruby_opts, "-X-C"
set :asset_env, "RAILS_GROUPS=assets EXECJS_RUNTIME='Node'"
I have successfully managed to set up a VPS production server (Ubuntu 10.04 LTS) from the excellent Railscasts episode. It's an "internal" server (not live yet), so I've been building my app locally and doing a cap deploy at regular intervals to check that things are running smoothly.
However, what I'd like to do now is delete all records on the production server (as I've just been testing stuff) - that is, to start with a completely empty database for when the site actually goes live to the public.
Obviously, I can do this locally by running something like rake db:reset, but how do I do this on the production server? Should I be adding some code to my deploy.rb file?
I'm a bit of a noob at this, but I've been unable to find anything via a Google search.
** EDIT ** Oh, and obviously this is a one-off - once things go live, I'll remove any code which deletes records!
You can ssh into the server and run any rake command from the application directory. You could create a Capistrano task just to run this one rake task, but since this task is obscenely dangerous with any real system I would not recommend it. The last thing you would want is to accidentally run it.