I am running CUPS V1.3.11 on Fedora 11.
I have a number of printers set up on the system that all work fine.
Some of these printers are on another system and I've mapped them using lpd. For example:

lpd://<ip address of server>/queueName

I noticed the other day that one of these servers had gone down and so the printer attached to it was unavailable. The CUPS error_log reported that the printer was not available and it was going to retry. So far, so good as I have the job-retry option set.

However, the JobRetryInterval is set to 60 seconds with a JobRetryLimit of 1440. This should retry for a full 24 hours before giving up.

The error_log doesn't say this though. It says:

network host '<ip address>' is busy; will retry in 5 seconds
and then after 5 seconds
network host '<ip address>' is busy; will retry in 10 seconds
and then after 10 seconds
network host '<ip address>' is busy; will retry in 15 seconds
and then after 15 seconds
network host '<ip address>' is busy; will retry in 20 seconds
and then after 20 seconds
network host '<ip address>' is busy; will retry in 25 seconds
and then after 25 seconds
network host '<ip address>' is busy; will retry in 30 seconds
and then after 30 seconds
network host '<ip address>' is busy; will retry in 30 seconds
and then after another 30 seconds
network host '<ip address>' is busy; will retry in 30 seconds

and so on.

Does anyone know why CUPS is choosing to ignore the JobRetryInterval and where this incremental interval behaviour is coming from ?