I think what's happened is that development is pretty much at a
standstill. I'm not saying "Dammit Vikas! Get this thing going again!",
just observing...
Anyhow, in our case we're running a few versions behind current, and the
local mods are so heavy I couldn't even imagine upgrading. In talking to
other people that use Nocol, I've found the same thing. The changes are
broad and somewhat hackish, but they work, and have the side-effect of
locking you into what you've got.
I think there are a few things that would make a new nocol better would
be:
-drop the 'hostmon' perl client and use something like ucd-snmp to gather
host stats. Include any extra agent functions that don't come with it but
are useful.
-db support really seems like a wonderful idea. Don't forget something to
check that your db is responding as expected.
-more snmp monitoring stuff, with configs for common equipment
-since you're already looking at the status of interfaces, why not get a
count and add a mrtg-like (but dynamically generate) usage graph for the
interfaces.
-hostmon-like alarms on interfaces when a threshold is reached.
-standardize on one pager package to allow for easier 'out of the box'
setup. Qpage has been great for us (www.qpage.org).
That's all. The database idea is wonderful...
Charles
--
=-----------------= =
| Charles Sprickman Internet Channel |
| INCH System Administration Team (212)243-5200 |
| spork@inch.com access@inch.com |
= =----------------=
On Wed, 13 Oct 1999, Velocet wrote:
> then Roger Burton West's all..
> >
> > On or about Tue, Oct 12, 1999 at 03:06:43PM -0400, Jonathan A. Zdziarski typed:
> > >lots of folks have switched to the commercial software 'netcool' which
> > >runs about $100,000 or so. it's much nicer, but obviously nocol has its
> > >own value in being free.
> >
> > "Much nicer" is an interesting way of putting it. "A total waste of
> > money" would be more accurate - the probes don't work, all the IP-
> > based probes _insist on_ DNS rather than IP numbers - so useful if
> > your nameserver falls over, all of a sudden all the probes fail -
> > and you need a high-end system to run it on.
> >
> > I've looked at a lot of different monitoring packages, and they're
> > all horribly expensive (or freeware). I haven't yet seen one that
> > worked as well as NOCOL.
>
> If nocol is so good, why is the support so lacking? Or am I just blind?
> Is there activity on this list beyond people jsut trying to get things
> going and a few helpful people who maintain their monitors diligently
> while the core package gets more and more outdated? Is there an effort
> to release a new version or add some of the features I and 100 other people
> are looking for?
>
> Is someone attempting to kick start it? Are the original author(s) still
> working on it, or will it be taken over (or worse, will a code fork be
> the only thing to kickstart the authors, resulting in two development
> paths)?
>
> > >> Has anyone got nocol doing hierarchical conditions, so that when a link
> > >> is down you dont also get paged with SMTP, Web, Pop, Imap, Ftp and everything
> > >> else being down too?
> >
> > No... but on the other hand I have got a once-every-five-minutes cron
> > job that tells me all the system status changes in that time.
>
> Hmm. I wonder if we can fit these messages into a single 100 char alpha
> page. ;) This seems to be the way to go with paging - collect all
> information until some cutoff time (every 5 min) and then send a page
> of distilled most-critical information.
>
> You got this pager interface code handy?
>
> Does anyone have a nice generic paging scheduler with easy to setup
> weekly schedules for people on call?
>
> /kc
>
|