Re: [Netdisco] Setting macsuck, arpnip to run once a day

[Oliver G. <ol...@cp...>]

On 2015-01-02 20:46, Simon Hobson wrote:
> There's an argument for having a "free running" task which sends
> queries spread out - applying a maximum rate for the queries, and a
> minimum interval between updates for a single host. It's what Nagios
> does for it's checks - though it's significantly more programming 
> work
> than just firing off a task to update all devices each time it's run.

Yeah I think someone did ask for a "continuous poller" before... ah yes 
here's the ticket if you want to add some voice:

https://sourceforge.net/p/netdisco/nd2-features/14/

For further general interest...

What's perhaps also of note is that we do know of many installations 
with large numbers (hundreds to thousands) of devices who do manage 
hourly polling. I think it has as much to do with the horsepower at the 
polling end, and the network device CPUs themselves (slow devices tie up 
the pollers).

Pollers (i.e. netdisco-daemon) can be run simultaneously on multiple 
servers as well as of course multiple cores on one system. Also giving 
the DB more resources helps (and using pgtune script on the PostgreSQL 
config).

regards,
oliver.