Multiple levels of "instances"?
David Wolfskill <david.wolfskill <at> wiline.com>
2004-11-04 18:06:03 GMT
Cricket newbie here; using 1.0.5 on FreeBSD 4.10-STABLE (via FreeBSD
ports), only for the last 3 days or so.
Cricket looks quite promising, but I'm getting a bit confused as to how
to set up the monitoring for the specific class of devices for which I'm
investigating Cricket, and I'd appreciate some guidance. [Yes, I have
searched through the archives to some extent, as well as read the bulk
of the reference manual & other forms of docs.]
The issue is that the devices I'm trying to monitor have something
akin to "multiple interfaces" at more than one level, and the number
of those "multiple instances" can change over time.
More specifically, I'm dealing with Aperto PacketWave 1000 Base
Station Units (BSUs). Each such BSU has one Ethernet NIC and from
0 (well, 0 would not be useful, but...) to 6 "Wireless Subsystems"
("WSSs" -- radios, really). Each WSS, in turn, may be connected
to some number of complementary "Subscriber Units" ("SUs" --
basically, CPE gear; one side radio and the other is Ethernet). (A
given WSS is, in our usage, typically attached to a sector antenna, and
used as part of a "point-to-multipoint" configuration.)
The objective is to monitor various health & signal quality metrics, as
well as traffic, for the various components of the system. This
includes the BSU itself, each WSS, as well as each SU.
(If we get this far, we also would probably want to go a bit further:
we will typically feed multiple customers from a single SU, and being
able to break those out would probably be useful at some point in the
future. Still, that's more a "desired direction" than an immediate
The number of WSSs connected to a given BSU is admittedly relatively
constant, so I could implement this part as separate targets. But the
only difference between looking at an OID for WSS1 vs. the corresponding
OID for WSS2 is that the first ends in .1, while the second ends in .2.
So I think I'd prefer to "lump things together" if that's feasible.
[Indeed, I've managed to get to this point so far -- though it's rather
imperfect, at best. The "Index" is a little weird, but seems to work.
And I need to improve my data-representation skills. :-}]
But I'm a bit stumped when I get to the SUs: I don't see how I can use
"instance mapping" for the SUs if I'm also using it for the WSSs.
If someone has an example that illustrates a way of doing this type
of thing, and wouldn't mind sharing (a sanitized version of) it,
or could point me at where such an approach is documented, I think
that would be ideal.
In any case, if I get this working, I plan to contribute it back, both
to Cricket and to Aperto. :-}
David H. Wolfskill 510 377-7226 david.wolfskill <at> wiline.com
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.