Unsendable alerts are logged on every alertengine queue check

[lunkwill42]

Scenario observed under NAV 4.8.5:

A user has a profile with a Jabber alert address. However, Jabber is not configured on this NAV server. The Jabber dispatcher will log one exception as /alertengine/ starts, since it cannot read the "required" configuration from smsd.conf. In turn, the dispatcher will be blacklisted, since it fails, but alertengine will keep queueing all new alerts for this user to this address.

Not only that, but every queued alert will be logged on every queue run, which massively increases alertengine's log output. In the abovementioned scenario, the user had 642 queued alerts, which led to 642 log lines being emitted every 30 seconds, quickly filling the log partition.

A potential solution might be to differentiate between a blacklisted and non-configured dispatcher, so that alerts addressed to non-configured address types can be discarded immediately.

A fruther improvement would be to deny a user to use alert addresses of non-enabled types.