Re: mrtg and squid

On Tue, 22 Dec 1998, Martin Hamilton wrote:

> I think the problem is that MRTG (recent versions, at least) always
> wants to query for sysName and sysUptime, whatever OIDs you ask it to
> graph, and whatever the setting for routerUptime.

Oh dear, I have to admit that you're right. Looking in the code I can
now see your point. It does indeed attempt to retrieve (and ignore)
the sysUptime and sysName OIDs from the actual "interfaces", even
though it's been told to fetch (and will in fact use) the ones
specified in routerUptime directive.

I now realise that in my situation, where, as I said, my MRTG is
actually graphing my multicast router on port 9161, these OIDs are
going to produce a (not very enlightening) result, which is ignored.

You say that squid behaves differently, and leads to failure.

I was convinced, from seeing in my graphcs the meaningful uptime and
system name that I was pointing it at, that everything was working as
intended, and I stubbornly insisted that it would work equally well
for squid.

> It queries on the
> same port you're monitoring in the Target specification.

It's actually querying both. I can see that, when debugging is turned
on.

> This should really be fixed by MRTG itself, though...

Agreed. Will I write to the author to explain the issue?

I'd like to thank you for your patience and for putting your finger
on the specific problem. Without your targetted(!) advice, I think I
would still be stubbornly insisting that it must work.

Humble apologies to all.
Received on Tue Dec 22 1998 - 10:47:23 MST