Quote from: DiceBitcoin on September 20, 2014, 11:15:40 PM

Quote from: dooglus on September 20, 2014, 09:21:32 PM

Quote from: suchmoon on September 20, 2014, 08:52:39 PM

There is one aspect of that story that's still bothering me (well there is more than one TBH, but I'm trying to pretend now that it's true). They said the the rogue employee "had to chose manually which player to cheat". How did he do that if he didn't have access to the production database? Some kind of a backdoor in the UI? All we've seen was two or three lines of code that don't really explain much. I think at the very least DB should have published the whole commit. This is one of those things that would have gone a long way towards credibility.

Good point.

The three line screenshot didn't really show anything. Could we see the whole diff he submitted? The condition for when to apply the nonce-skip would be interesting. I too wonder how the rogue employee was able to chose manually which players to cheat when he didn't have access to the db.

To answer your question there was no backdoor on the UI. Besides the "main" website we also have an application in place that we use to do simple day-to-day operations (such as reseting users password, process manual withdraws etc), an 'admin' application if you like.

In this application the are only stuff that anyone can view (no secret stuff lying around) and do. One of those is that one could view / edit a JSON field on the user that we use primarily for storing meta information (for you techies take a look here: http://www.postgresql.org/docs/9.4/static/datatype-json.html) such as last-login, how much time he is active etc, nothing important. He used this schemaless column to store the data he wanted in order to persist the conditions that when met the skipping happened.

Doog, as for the diff I will post it later

Here is the diff for the malicious code commit, for what it's worth http://pastebin.com/anXZmNM6