The expression “Race Condition” is used to refer to a state in which multiple processes of a system execute in parallel, whereby a strict adherence to a particular order of steps is expected, but cannot be guaranteed, due to a lack of coordination between the entities. This oftentimes produces unpredictable behaviour and incorrect outcomes.
Example
Two simultaneous processes need to increment the same value which can only be retrieved or stored to in one operation (step).
| Step | Process A | Shared value | Process B |
| 1 | Retrieve [Internal value: 0] | 0 | Retrieve [Internal Value: 0] |
| 2 | Increment [Internal value: 1] | 0 | Increment [Internal value: 1] |
| 3 | Store | 1 | Store |
As you can see due to neither process knowing that the shared value was being manipulated elsewhere, only one of the operations was conducted successful with the other’s work being overwritten. This may seem to be nothing more than a cool though experiment, but Imagine processes A and B represent concert-goers booking tickets to see their favorite band, and the shared value is the number of available seats. Now say there’s only one ticket left, but they book simultaneously or worse, the same happens to dozens of pairs.
Remediation
To fix this bug, a data synchronization mechanism such as a mutex lock (see example below) must be introduced.
| Step | Process A | Shared value | Process B |
| 1 | Lock value against outside influences | 0 | |
| 2 | Retrieve [Internal value: 0] | 0 | Attempt to lock value => Failed |
| 3 | Increment [Internal value: 1] | 0 | Wait for release of lock… |
| 4 | Release lock | 1 | Wait for release of lock… |
| 5 | 1 | Lock value | |
| 6 | 1 | Retrieve [Internal value: 1] | |
| 7 | 1 | Increment [Internal value: 2] | |
| 8 | 2 | Store | |
| 9 | 2 | Release lock |
EN 
DE