Need some guidance on interpret and query table

[owner66]

Hi,
I have cacti up and running, and am able to graph values for my wireless nodes using single OID using data templates.

I am looking into using queries for more information, but I am having issues interpreting the walk on the wifiRadioStatsEntry table.
I read about tables in the turorials, but I seem to have either a unique requirement, or a total mis-understanding of tables.

The wifiRadioStatsEntry table represents the stats for the connection between 2 radios (station and AP). In other words, when i poll 142, i get the information for 142 (station) AND 147 (AP) (see attached inventory doc).
Furthermore, it seems that i cannot rely on the 1st entry in the table to be the station or the AP.

MAC 00:06:5A:00:02:AC connects wireless to 00:06:5A:00:33:FC
MAC 00:06:5A:01:EA:23 connects wireless to 00:06:5A:00:33:FC

When i poll .142, oxFC (the station) is listed 1st, but when i poll .143, oxAC (the AP) is listed 1st.

Given this, how do i tell cacti which value to poll? I CAN get which MAC addresse(s) are stations from another OID, so I can provide that as a value to use to determine which row in the table i need.

Where to go from here? suggestions?
Am I not intrepreting the contents of the table correctly?

[gandalf]

The MIB should explain, how the index to that table is composed. There is no general way how to do this; manufacturers may work this out their own way.
It looks like this index is composed of two pairs of addresses; whichever meaning those might have.
R.

[owner66]

I'm not getting alot of help from manufacturer on this.
I can poll and manipulate the values using a php script and put the values i need in a mysql table.
Can cacti be used to get snmp values from a mysql database? Not database stats, an actual 'walk' of the table?

[gandalf]

Can cacti be used to get snmp values from a mysql database? Not database stats, an actual 'walk' of the table?

Do you mean "snmp values (=data)" or "snmp OIDs (to fetch data)"?
But nevertheless, by writing a script that does exactly what you want and that finally spits out the required raw data, you can cope with both approaches
R.

[owner66]

I would use the script to poll the oids, and get the data. then store the data with a key(oid) or other identifier in a mysql table in rows.
Then i would use cacti to graph the data from this mysql data source.
So cacti CAN retrieve data from mysql as though it were a host?
Sorry for dragging this out, i just really like the cacti interface, the ability to create users, ease of use etc.
I just am thinking out loud about how to accomplish this.

[gandalf]

You can do so. But I do not understand, why you'd like to move the data through mysql first.
And yes, you will have do define a "pseudo-host" that "polls" the mysql table by using your specific script
R.

[owner66]

When i look at SNMP examples for say a network switch, i can see how an index would allow polling things like interfaces. A switch had 12 ports, each port is in the table based on an index. Each and every time, it is the same.

The wifi system that i am polling is made to automatically form a self-healing, self-configuring mesh of nodes. Their common goal is to determine a least cost path to a gateway(s). Poll a node 1 time and there may be 1 entry in the table, next time there may be 2 or more.

In the screenshots, i am showing a few nodes. When i walked them, it seemed to me that one radio listed itself first, and its peer second. The other radio listed it's peer first and itself second.

I don't see how i could use an index and tables when i am not sure of order or count.

my thought was to use an intermediate step, gather the data, put it into a dbase to filter and order it (or sort in the script), and present it to the poller.

Am i just reading the screenshots i provided wrong and over analyzing?

I do appreciate your help with this.

[gandalf]

Of course you may do so.
There's the CPU Data Query, which kind of "re-indexes" the data presented by SNMP as the index changes on reboot. This may serve as a "small" solution.
But if your post-processing is more complicated, you may of course act as described. Finally, You will have to write a script that pulls out the data from mysql.
R.