Right now /var is cleaned up periodically. In most cases, the frequency is fine, but if there's another bug that begins filling the folder quickly it can leave the device in an unbootable state (see bug 664659).

There are a couple obvious options to mitigate this.

Option A:
During normal operation detect if more than x% of /var is used (maybe 75%) and automatically trigger a cleanup if the threshold is crossed.

Option B:
Detect a full /var at boot time and clean up then.

There are probably a dozen other non-obvious options. The main goal is that a bit of runaway logging shouldn't be able to easily leave a machine in a non-bootable state.