Comment 2 for bug 1214274

- case 1 failure. We forked select and update queries, and run xtrabackup.
  xtrabackup started before forked processes started.
  * possible solution is to introduce synchronization in test case between
    forked queries and main execution thread
 * more simple solution is to increase timeouts in xtrabackup parameters to
   very large value
- case 1 failure. Mysql server has gone away query killing has been finished.
  Looks like xtrabackup killed it's own connection.
  * possible solution is to check if connection belongs to xtrabackup and don't
    kill it.

Both failures are case 1. However 2 may happen not only for case 1.
I propose to proceed with simple solution for case 1. Case 2 shouldn't be hard to implement either.