Came across this whilst looking at #685633 but thought that it deserved a separate bug.
If a cron function causes an exception for whatever reason, then the cron will stop, but the lock isn't removed. The job then isn't tried again for another 24 hours.
Whilst this is possibly correct in some fashions -- e.g. broken crons aren't run endlessly to no point, some crons may cause an exception once in a while which isn't a major issue but shouldn't mean that the cron doesn't take place for another 24 hours.
For example, network issues will probably cause the cron_check_for_updates cron to fail, but this could be a transient issue.
Not sure whether it's best to catch the errors and alert an administrator but still keep the lock in place (to prevent the admin from being spammed), or to catch the errors and assume that they may be transient and try again, or some other solution...
also related to bug 776955