Creating temporary files in Heroku

Question

I have an app hosted @ Heroku. The app depends on some feeds which are fetched using a socket listener. The socket listener gets one line of XML per second. Once I detect th

Answer 1

With the newer Heroku-16 stack, you are able to write to both the root and to /tmp

Try writing to root with

f = File.new("filename.txt", 'w')
f << "hi there"
f.close

Dir.entries(Dir.pwd) # see your newly created file

Or to /tmp with

f = File.new("tmp/filename.txt", 'w')
f << "hi there"
f.close

Dir.entries(Dir.pwd.to_s + ("/tmp"))

You will see your new file among those listed in both cases

Also try running heroku restart to see your newly created files disappear! This is expected, as heroku storage is ephemeral (will be deleted when the app restarts) - so don't rely on it for anything more than (very) temporary storage

Answer 2

The documentation on Heroku's read-only file system explains that you can use #{RAILS_ROOT}/tmp but doesn't give any examples for generating a temporary filename. This will guarantee uniqueness of the filename:

prefix = 'mydata'
suffix = '.xml'
Tempfile.new [prefix, suffix], "#{Rails.root}/tmp"

Note that "there is no guarantee that this file will be there on subsequent requests (although it might be), so this should not be used for any kind of permanent storage."

Answer 3

You may be able to use the #{RAILS_ROOT}/tmp/ directory or Rails.root.join('tmp').to_s:

Aspen & Bamboo
[...]
There are two directories that are writeable: ./tmp and ./log (under your application root).
[...]

Cedar
Cedar offers an ephemeral writeable filesystem. You can write out to disk anywhere you would like. Your changes will be lost on dyno restart and spin-up.

RAILS_ROOT is for older Rails versions, Rails.root is for newer versions.

You can't depend on anything surviving across requests of course, there's no guarantee that you'll even be working with the same dyno.

As long as you stay within the same process or request, Rails.root.join('tmp') should be usable. If you need the temporary data to survive across requests or processes then you're better off using something else (such as MongoDB or PostgreSQL) as a collecting ground for your data on its way to S3.

---

Thanks to Benjamin Wheeler for the heads up about the RAILS_ROOT to Rails.root change.