[CentOS] SNMP monitoring options

Wed Apr 13 01:30:53 UTC 2011
aurfalien at gmail.com <aurfalien at gmail.com>

On Apr 12, 2011, at 6:23 PM, Bob Hepple wrote:

> On Tue, 12 Apr 2011 12:53:41 -0700
> aurfalien at gmail.com wrote:
>> Hi all,
>> Been using Cacti for monitoring various things like, system disk/mem/
>> proc, network usage, router usage etc...
>> While its been fun, the graphs are just unruly.
>> Was looking an OpsView (the free version), wondering what your
>> experience with this type of trend/heuristic analysis has been and
>> what what you like.
>> And of course thoughts on OpsView.
> My 5 penn'o'rth ...
> We looked at a number of monitoring systems before closing on
> (community version) Opsview. The main other contender we looked at was
> zenoss - I have to admit I was biased due to previous use of and  
> liking
> of nagios.
> What we liked in Opsview were:
> based on nagios - solid pedigree, good technology, our own previous
> experience, lots of plugins built-in or available, in extremis you can
> look at the nice ascii configuration files and see what's going on.  
> You
> also have an escape route to nagios if opsview disappears (but they
> appear to be thriving AFAIK).
> easy to extend to custom tests/monitors using eg ssh scripts
> data sets are right in front of you - it would be easy to grab
> your data and run with it, if you had to (not that I've done much with
> it, but it's nice to know that your data is not held hostage in some
> binary silo).
> very light feel - I mean it's light on resources both on the testing
> machine and on the targets. It would probably scale up well (we only
> monitor about a dozen or so systems).
> pointy-click - so there is the (remote) possibility that I could lob
> this off onto someone else! The graphical i/f is rational - unlike  
> some
> others eg zenoss which I just couldn't get my poor old head around _at
> all_!!!
> opsview people and community are helpful, positive, approachable etc
> etc Community project is keenly supported and not just poor-cousin to
> paid-for product. Just like this newsgroup - if you post a message
> you're very likely to get someone pipe up with a helpful reply.
> As for trending/heuristics - the graphs are good enough for us as-is
> and the knowledge that you can plug-out a feed to your own
> datastore/analysis is comforting.
> The bad?
> nothing really. Well, twist my arm - the web i/f can be a bit  
> ponderous
> and there's a couple of gotchas that you just have to know about eg  
> you
> can make all the changes you like, but nothing actually takes effect
> until you find the configuration page and press the reload button.
> Also, new monitors need two re-loads before the graphs appear. I'm  
> just
> mentioning them to illustrate how trivial my gripes are.
> Hope this helps ...

Thanks Bob, great post.

Looking forward to Opsview.

- aurf