<AuthBy FAILUREPOLICY>
allows you to act on
repeated failures. The current implementation monitors usernames and only
considers those failures that indicate bad password as the failure
reason.
Two policers are currently implemented. One counts consecutive failures
and the other counts cumulative failures. Both policers implement separate
failure counters, policy violation time and windowing. Windowing is
optional and allows counters to be reset after the configured time window
has passed. Counters are not incremented when a policy violation is
active. If a violation is not active when a failure occurs, both
consecutive counter and cumulative counter are currently incremented.
Windowing is typically used with the cumulative policer. For example,
when cumulative failure threshold is set to 200 and windows size to 12
hours, the counter is reset every 12 hours. Window time is based on Unix
timestamp and is not relative to Radiator process startup.
Counters are kept in Radiator process memory. If multiple Radiator
instances need to share the counters or they need to persist across
process restarts and reloads, see
Section 3.103. <AuthBy SQLFAILUREPOLICY>
Important
See
goodies/failurepolicy.cfg
for an example
configuration. Currently a hook is needed to monitor failures and maintain
counter history.