I've installed the new SNMP package (snmpd-opkg) on a node (a Mikrotik, if that makes any difference).
I copied /etc/config/snmpd-opkg to /etc/config/snmpd and removed snmpd-opkg
On inspection it looks like the new /etc/config/snmpd config file should work right out of the box, unless I missed something.
Then I did
./etc/init.d/snmpd enable
./etc/init.d/snmpd start
It returns to the prompt with no error. However, doing an snmpwalk of the node ( snmpwalk -v1 -c public <nodename> system ) times out, and doing ps | grep snmp reveals that the daemon isn't running.
AM I doing something wrong?
Thanks.
Orv W6BI
I copied /etc/config/snmpd-opkg to /etc/config/snmpd and removed snmpd-opkg
On inspection it looks like the new /etc/config/snmpd config file should work right out of the box, unless I missed something.
Then I did
./etc/init.d/snmpd enable
./etc/init.d/snmpd start
It returns to the prompt with no error. However, doing an snmpwalk of the node ( snmpwalk -v1 -c public <nodename> system ) times out, and doing ps | grep snmp reveals that the daemon isn't running.
AM I doing something wrong?
Thanks.
Orv W6BI
http://downloads.arednmesh.org/releases/3/18/3.18.9.0/packages/mips_24kc...
It's not unique to the model #, rather to all mips_24kc cpu's (which includes ubnt, tplink, and mikrotik). Can't think of a reason why other models would not find it at the same location. If you are doing the nightly build, make sure you have the current build #. If the firmware is an older version, maybe snmp is installed already?
Joe AE6XE
Fired up Chrome, which I never use on the mesh network. Same negative results.
I'll keep researching it.
Installed snmp packages on the latest nightly build and was exploring the available content. There are almost 900 different values in the MIB tree, but only a few of them provide useful information -- mainly a subset of the Interface metrics as well as a couple of the Host Resources metrics. It would be quite a bit of effort to add node/RF info to the existing agent code. I was hoping for some RF metrics, so created a separate poller that grabs RSSI and SNR (just for now) and populates the NMS. I'm attaching a screenshot showing some of the NMS graphs so far. Next steps: automate node discovery in the NMS, and supplement with more node information (probably from olsr).
Joe AE6XE
Greg W2GMD,
I've been using Zabbix but you can use any "historian" to visualize node performance. I use Zabbix because it's easy for an external polling program to insert values into their database, and after that everything happens auto-magically. I'm really not much of a programmer so I'm just using a shell script, but you can use any programming language you're comfortable with. Currently my poller "screen scrapes" the values from the Node Status page, but screen scraping is not a very reliable approach. The right approach is to use the sysinfo API which will remain consistent into the future. The disadvantage is that nodes with an api_version less than 1.7 won't have the RF info. Values are returned as JSON, so your favorite JSON parser can grab the values and zabbix_sender inserts them into Zabbix.
One important comment here: Be careful which nodes you poll, how many metrics you query, and how often you retrieve them. It's best to limit polling to only those critical backbone or relay nodes that form the "spine" of your network. Also poll those selected nodes using a device that sits directly on that spine. Make sure you ask permission from the node owners before you start polling them.