Capistrano deploy code from bitbucket repository gets "authorization failed" - mercurial

Question

I'm trying to deploy code to an Amazon EC2 instance using Capistrano, a private bitbucket repository, and mercurial.
I successfully ran cap deploy:setup from my development box.
Then I tried to run cap deploy:cold. I was prompted to enter the bitbucket password, which I did correctly, but I got "authorization failed" error message. As a result, the machine to which I was trying to deploy the code could not "clone" the code from the bitbucket repository.
This is what happened in details:
% cap deploy:cold
triggering load callbacks
* 2013-01-10 18:44:07 executing 'production'
triggering start callbacks for 'deploy:cold'
* 2013-01-10 18:44:07 executing 'multistage:ensure'
* 2013-01-10 18:44:07 executing 'deploy:cold'
* 2013-01-10 18:44:07 executing 'deploy:update'
** transaction: start
* 2013-01-10 18:44:07 executing 'deploy:update_code'
updating the cached checkout on all servers
executing locally: "hg log --verbose -r default --template \"{node|short}\""
command finished in 797ms
* executing "if -d /home/ec2-user/my_project/shared/cached-copy ; then hg pull --verbose --repository /home/ec2-user/my_project/shared/cached-copy https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project && hg update --verbose --repository /home/ec2-user/my_project/shared/cached-copy --clean 2ec0193d60ff; else hg clone --verbose --noupdate https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project /home/ec2-user/my_project/shared/cached-copy && hg update --verbose --repository /home/ec2-user/my_project/shared/cached-copy --clean 2ec0193d60ff; fi"
servers: "ec2-123-456-78-999.compute-1.amazonaws.com"
ec2-123-456-78-999.compute-1.amazonaws.com executing command
** ec2-123-456-78-999.compute-1.amazonaws.com :: out http authorization required
** ec2-123-456-78-999.compute-1.amazonaws.com :: out realm: Bitbucket.org HTTP
** ec2-123-456-78-999.compute-1.amazonaws.com :: out user: bitbucket_user_id
** ec2-123-456-78-999.compute-1.amazonaws.com :: out password:
hg password:
** ec2-123-456-78-999.compute-1.amazonaws.com :: out
** ec2-123-456-78-999.compute-1.amazonaws.com :: out abort: authorization failed
command finished in 5410ms
*** deploy:update_code rolling back
* executing "rm -rf /home/ec2-user/my_project/releases/20130110184415; true"
servers: "ec2-123-456-78-999.compute-1.amazonaws.com"
ec2-123-456-78-999.compute-1.amazonaws.com executing command
command finished in 431ms
failed: "sh -c 'if -d /home/ec2-user/my_project/shared/cached-copy ; then hg pull --verbose --repository /home/ec2-user/my_project/shared/cached-copy https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project && hg update --verbose --repository /home/ec2-user/my_project/shared/cached-copy --clean 2ec0193d60ff; else hg clone --verbose --noupdate https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project /home/ec2-user/my_project/shared/cached-copy && hg update --verbose --repository /home/ec2-user/my_project/shared/cached-copy --clean 2ec0193d60ff; fi'" on ec2-123-456-78-999.compute-1.amazonaws.com
This is my Capistrano deploy.rb file
set :application, "my_project"
set :repository, "https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project"
set :keep_releases, 4
set :scm, :mercurial
set :scm_username, "bitbucket_user_id"
default_run_options:pty = true
set :scm_prefer_prompt, :true
set :scm_verbose, :true
set :deploy_to, "/home/ec2-user/#{application}"
set :deploy_via, :remote_cache
set :user, "ec2-user"
set :use_sudo, false
namespace :deploy do
desc "Gracefully restarting unicorn"
task :restart, :roles => :app do
run "sudo /etc/init.d/start_script upgrade"
end
task :start, :roles => :app do
run "sudo /etc/init.d/start_script start"
end
task :stop, :roles => :app do
run "sudo /etc/init.d/start_script stop"
end
task :link, :roles => :app do
run "mkdir -p #{shared_path}/images"
run "ln -nfs #{shared_path}/images #{release_path}/tmp/images"
end
end
I tried this manually on the command line
ssh ec2-user#ec2-123-456-78-999.compute-1.amazonaws.com "hg clone --verbose --noupdate https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project /home/ec2-user/my_project/shared/cached-copy"
and I got (no prompt for bitbucket password)
abort: http authorization required
I tried
ssh -t ec2-user#ec2-123-456-78-999.compute-1.amazonaws.com "hg clone --verbose --noupdate https://bitbucket_user_id#bitbucket.org/bitbucket_user_id/my_project /home/ec2-user/my_project/shared/cached-copy"
and it worked.
I believe Capistrano is using ssh (with pseudo tty turned on by this line in my deploy.rb)
default_run_options[:pty] = true
So it should work, but why it isn't working with Capistrano?
Can someone help with this please?
Thanks much!

Answer 1

I was having a similar issue to this and solved it by looking at this question: Capistrano deploy failed after moving to bitbucket.org
An answerer there suggested that the person check that they have made deployment keys in BitBucket. In your BitBucket repo, click your settings wheel and in the left hand sidebar click "Deployment Keys": Use deployment keys to gain read-only access to this repository.
If you don't know how to set up SSH keys, read BitBucket's guide here: https://confluence.atlassian.com/pages/viewpage.action?pageId=270827678
Hope this helps!