You can clone with
HTTPS or Subversion.
We see that when the server return some error, specify 503 error code- "Service unavailable", the completion block is called and not the fail block.
Isn't it should call the failure block, or the failure block should only called when the request itself failed, and I should check the response code on myself?
you should check the response code yourself :D
That's what I do now, but I think that all error handling should be in the failure block, don't you think?