Capistrano / Mercurial / tar "time stamp in the future" Message

Question

I'm experiencing an intermittent problem when deploying my rails app with capistrano. After the .gz is uploaded to the server, the log shows an error coming from tar, which says that the time stamp is in the future (see output below). This message repeats for what seems to be every file in the app.

I compared the server time against the numeric name of the .tar.gz file, and it's indeed a few minutes ahead. But how does this correlate to the actual timestamp on the files? And how can I get my changes to deploy correctly?

** sftp upload /var/folders/lo/loIUAyGAHFWqSREiNHhm-E+++TI/-Tmp-/20110603143429.tar.gz -> /tmp/20110603143429.tar.gz
   [123.45.67.890] /tmp/20110603143429.tar.gz
   [123.45.67.890] done
 * sftp upload complete
 * executing "cd /home/blah/releases && tar xzf /tmp/20110603143429.tar.gz && rm /tmp/20110603143429.tar.gz"
   servers: ["123.45.67.890"]
   [123.45.67.890] executing command
** [out :: 123.45.67.890] tar: 20110603143429/.autotest: time stamp 2011-06-03 14:34:33 is 368.72042712 s in the future
** [out :: 123.45.67.890] tar: 20110603143429/.bundle: time stamp 2011-06-03 14:34:33 is 368.719540808 s in the future
** [out :: 123.45.67.890] tar: 20110603143429/.hgignore: time stamp 2011-06-03 14:34:33 is 368.719465444 s in the future
** [out :: 123.45.67.890] tar: 20110603143429/app: time stamp 2011-06-03 14:34:34 is 369.719382175 s in the future

 ** [out :: 123.45.67.890] tar: 20110603143429: time stamp 2011-06-03 14:34:49 is 383.369448435 s in the future
    command finished in 1616ms
  * executing `deploy:finalize_update'
  * executing "chmod -R g+w /home/blah/releases/20110603143429"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
    command finished in 193ms
  * executing "rm -rf /home/blah/releases/20110603143429/log /home/blah/releases/20110603143429/public/system /home/blah/releases/20110603143429/tmp/pids &&\\\n      mkdir -p /home/blah/releases/20110603143429/public &&\\\n      mkdir -p /home/blah/releases/20110603143429/tmp &&\\\n      ln -s /home/blah/shared/log /home/blah/releases/20110603143429/log &&\\\n      ln -s /home/blah/shared/system /home/blah/releases/20110603143429/public/system &&\\\n      ln -s /home/blah/shared/pids /home/blah/releases/20110603143429/tmp/pids"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
    command finished in 257ms
  * executing "find /home/blah/releases/20110603143429/public/images /home/blah/releases/20110603143429/public/stylesheets /home/blah/releases/20110603143429/public/javascripts -exec touch -t 201106031435.03 {} ';'; true"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
    command finished in 1911ms
    triggering after callbacks for `deploy:update_code'
  * executing `bundle:install'
  * executing "ls -x /home/blah/releases"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
    command finished in 140ms
  * executing "bundle install --gemfile /home/blah/releases/20110603143429/Gemfile --path /home/blah/shared/bundle --deployment --quiet --without development test"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
 ** [out :: 123.45.67.890] sh: bundle: not found
    command finished in 158ms
*** [deploy:update_code] rolling back
  * executing "rm -rf /home/blah/releases/20110603143429; true"
    servers: ["123.45.67.890"]
    [123.45.67.890] executing command
    command finished in 261ms
failed: "sh -c 'bundle install --gemfile /home/blah/releases/20110603143429/Gemfile --path /home/blah/shared/bundle --deployment --quiet --without development test'" on 123.45.67.890

Answer 1

It sounds like you have have clock-skew. All all the computers involved in the process -- those commiting, those sending, those receiving/deploying in sync?