FIX: under rare conditions saving a new draft could error temporarily

Under rare conditions due to bad HTTP timing and so on a draft could be
set at the exact same time from 2 unicorn workers.

When this happened and all the stars aligned, one of the sets would win
and the other would raise an error.

This transparently handles the situation without adding any cost to the
draft system.

The alternative is to add a distributed mutex, tricky DB transaction or handle
the error in the controller. However this seems like a reasonable way to
work around a pretty big edge case.

app/models/draft.rb

@@ -7,7 +7,7 @@ class Draft < ActiveRecord::Base
 
   class OutOfSequence < StandardError; end
 
-  def self.set(user, key, sequence, data, owner = nil)
+  def self.set(user, key, sequence, data, owner = nil, retry_not_unique: true)
     if SiteSetting.backup_drafts_to_pm_length > 0 && SiteSetting.backup_drafts_to_pm_length < data.length
       backup_draft(user, key, sequence, data)
     end
@@ -65,13 +65,24 @@ class Draft < ActiveRecord::Base
     elsif sequence != current_sequence
       raise Draft::OutOfSequence
     else
-      Draft.create!(
-        user_id: user.id,
-        draft_key: key,
-        data: data,
-        sequence: sequence,
-        owner: owner
-      )
+      begin
+        Draft.create!(
+          user_id: user.id,
+          draft_key: key,
+          data: data,
+          sequence: sequence,
+          owner: owner
+        )
+      rescue ActiveRecord::RecordNotUnique => e
+        # we need this to be fast and with minimal locking, in some cases we can have a race condition
+        # around 2 controller actions calling for draft creation at the exact same time
+        # to avoid complex locking and a distributed mutex, since this is so rare, simply add a single retry
+        if retry_not_unique
+          set(user, key, sequence, data, owenr, retry_not_unique: false)
+        else
+          raise e
+        end
+      end
     end
 
     sequence