On Fri, Jan 21, 2005 at 09:00:53AM -0800, Rex Buddenberg wrote:
> Dan,
>
> I'd turn the architectural question around just a bit.
>
> 1. [/proc -> inconsistent between releases/distributions]
>
> 2. [snmpd as a mitigator]
I, too, agree with this in principle... reading things out of /proc
could be bad, might require elevated priviledges, etc (and as Rex
stated, isn't necessarily consistent between different versions or
distributions of linux).
So, to mitigate this, you're forced back to the system utilities and
hostmon strives to deal with things that are fairly consistent between
most un*xes, and the snmpmon portion strives to move more things
directly in to snmp (though not everyone uses or even "likes" snmp on
their networks). For all the things in-between, there are plugins that
have been provided back to the project (such as apcmon, ntpmon, and a
wide variety of others).
So really, what I guess I'm trying to say is that it's a joint venture;
for items that have a spot on the system MIB, there's always that...
aside from that, there tend to be many other hooks back in to
nocol/snips. Perhaps Daniel wants to try to write a procfsmon... though
I'm guessing it'll be as tough (or worse) than hostmon to try to
stabilize?
Just my $0.02, FWIW...
Russell
--
Russell M. Van Tassell
russell at loosenut com
"I think it's more flexible because it'll give us more flexibility to
us" -- My PHB, 2004/03/30
|