[wellylug] Load average statistics dubious

[Daniel Pittman]

"C.T.F. Jansen" <frank.jansen at actrix.gen.nz> writes:

> I used to use a number of different unix OS's and the load average
> statistics on them is an artificial number and often with values too
> different with what is actually happening on the system. It was regarded
> as a joke and people warned not to use it.

Presumably the people warning you of this were scared by too many years of
people screaming because the load average showed a perfectly reasonable
number, but who had a religious belief about what it should be.

> top and ps generate worthwhile numbers.

Nonsense: top produces entirely misleading figures regarding memory use, since
it does not accurately report shared vs non-shared data.  Which, of course, is
a lesson in making sure you appropriately limit your statements to only the
aspect that you *meant* to include in the statement.

[...]

> I was particularly interested in iowait and similar monitoring. Its
> taken quite a while for the linux/unix community to catchup with the
> adequate mainframe perfomance monitoring software.

You sound surprised.  Unix has a decades old tradition of waiting thirty years
or more to adopt mainframe features; see the whole upswing of interest in
containers, paravirtualization, and virtualization, for example.


> The linux journal author preferred nagios. I installed and started setting
> it up. There's quite a lot to it. Beware that it will install apache and
> samba for you as well!

You mentioned Debian earlier, so I assume you mean that this is true on
Debian, at which point I am comfortable to say that this is ... misleading, at
the very least.

Nagios requires an HTTP server of some sort, and the Debian default is
Apache.  Without that you couldn't actually *view* the status of systems.

Only the samba client is actually required; Debian break this up so that you
can get only smbclient, not the entire server stack.


Oh, and you didn't discuss nagios2 vs nagios3 at all.

Um, and how does nagios have anything to do with the load average, with which
you opened this discussion?  It surely has no capacity to measure the same
values, or the values from top/ps which you claim to be more accurate...

        Daniel

-- 
✣ Daniel Pittman            ✉ daniel at rimspace.net            ☎ +61 401 155 707
               ♽ made with 100 percent post-consumer electrons