News for hackers. Stuff that matters.
Re:Scaling lessons learned at Dropbox, part 1 by christoph on Thursday July 12, 2012 @08:40PM (#4237626)
So, so true.

When the problem occurs, it's pretty quick for the guy who needs to fix it at 2am, to find where it exploded in the code base, while the original developer is (maybe) passed out in a bar somewhere.

Not much else matters. He could of just done :( x 10 and had the same result. The main thing is, it's easily traceable!

Re:Scaling lessons learned at Dropbox, part 1 by akent on Thursday July 12, 2012 @08:58PM (#4237692)
Sure, for fatal errors, random (greppable) strings aren't so bad, but the OP made it sound like his FFFFFFFFFFFUUUUUU message was expected all the time rather than an exception. If you're going to print something all the time in normal operation, make it meaningful.
Re:Scaling lessons learned at Dropbox, part 1 by aboodman on Friday July 13, 2012 @02:51AM (#4238553)
That isn't how I took it. I read it as: there was an error scenario that was clearly happening, which should have resulted in a particular log message. Except it wasn't, which meant that something was racing.