[u-u] Asset tracking...
colin.mc151 at gmail.com
Wed Jan 14 10:47:37 EST 2015
On Wed, Jan 14, 2015 at 9:49 AM, Eric Siegerman <pub08-uu at davor.org> wrote:
> On 01/13/2015 02:53 PM, Alan J Rosenthal wrote:
>> Those items which can't be derived automatically get added to
>> the inventory files described at the beginning of this message.
> I take you to mean that the static data lives only in the central
> (text) database on the hub box.
> One could put that static data on the machine it pertains to (in
> a file in /usr/local/etc, say), and have the cron job include it
> in the hourly email along with the dynamically generated stuff.
> That would make each machine the authoritative source for all of
> the asset data about it, rather than splitting the "source of
> truth" between the machine itself and the central database.
> As well, it would make it easier to maintain that info -- when
> you make change something about a box, just edit a local file on
> the box itself. That's a huge win if there are firewall issues
> in ssh'ing from the individual boxes into the hub; but even
> absent such issues, it might be psychologically easier (or at
> least, it would be for me, though your brain might work
> differently from mine :-) )
> Besides homebrew scripts for automating the data collection,
> there are also tools like cfengine, puppet, chef, pikt. They all
> have the advantage that once you've set one up, I'm sure you'll
> find other uses for it. So probably more effort up front before
> you get to the point of automating your first task (asset
> tracking in this case), but the second automated task becomes a
> lot easier; and once you have a whole suite of such tasks, the
> fact that they're all run under one tool should make life
> easier/saner. (I haven't used any of these tools, so can't offer
> any advice on choosing among them.)
A comment I heard from one of the managers was that he had built two
asset tracking systems at two previous employers and he was wasn't
interested in doing a third. So, looking for a more-or-less
off-the-shelf Linux-friendly solution.
The firm is already using "Puppet" to fire some configuration files
out to some servers, but I gather it isn't good enough for tracking
some details regarding all the servers... Thus my wondering about
"Collins" https://tumblr.github.io/collins/index.html which sounds
like it could do the trick (write some configuration files, enter some
data and done). Question, is there something "better" (for any of
several definitions of "better") than "Collins"? Does "Collins" have
some sort of hidden, lurking horror(s) that could cause my employer a
megaton of grief?
More information about the u-u