Cisco Bandwidth ??
I"m trying to setup bandwidth monitoring for cisco switches and I'm having a really hard time. first of all....using the build in interface monitors seem to only graph the rate of change and not give me any current bandwidth values.
Also, that only works for specific interfaces and what I would really like to do is just graph total inbound and outbound bandwidth for the entire switch.
I get all data when I snmpwalk from the command line but can't seem to get things going in netmrg. Is there any documentation someone has out there explaing how to setup monitoring for cisco devices. I would really appreciate it and it would save me from alot of headaches.
[quotea09077f2bd="Infinity"]I"m trying to setup bandwidth monitoring for cisco switches and I'm having a really hard time. first of all....using the build in interface monitors seem to only graph the rate of change and not give me any current bandwidth values.[/quotea09077f2bd]
The provided graphs do graph the rate of change (ie, bytes/second). I'm not sure what you mean by "current" bandwidth values though. The counters the switch provides are sums of the traffic passing through the ports, so a graph of those values as they are would result in a graph that continually ascends until it wraps back to zero, which isn't normally very useful. If you mean the average values that the switch shows for traffic (via a "show interface" on the cisco), these aren't very useful for graphing purposes (as using rate-of-change with the counters provides higher resolution and more accurate graphs).
[quotea09077f2bd="Infinity"]Also, that only works for specific interfaces and what I would really like to do is just graph total inbound and outbound bandwidth for the entire switch.
I'm not aware of Ciscos providing this info via SNMP, so the best you could do would be make a stack graph of the individual interfaces. It would be tedious, but would get you what you want.
[quotea09077f2bd="Infinity"]I get all data when I snmpwalk from the command line but can't seem to get things going in netmrg. Is there any documentation someone has out there explaing how to setup monitoring for cisco devices. I would really appreciate it and it would save me from alot of headaches.[/quotea09077f2bd]
Could you post the info from snmpwalk that you're interested in graphing? That might help us see what you're trying to do. I monitor a lot of Ciscos, but for switches I seldom need anything beyond the regular interface template applied to all interfaces.
Hopefully we can help... and good luck.
Okay, another employee at my job has setup RTG to graph the traffic crossing the switch, and he's able to show traffic for each interface. I don't know how because when I do it with the prebuilt graphs...I get what I said before...rate of change. What I'm looking for is how much traffic is crossintg that interface at the time it's polled which is what the RTG graphs are showing.
I'll ask him what oid he's using to give you a better idea...I'm sure from what I understand that netmrg can do the same thing but I just don't know how. Once I get the oid, I'll reply and maybe we can get somewhere with this.
I must tell you...I like netmrg and see alot of future for this application...I really think it has potential for being the best monitoring software out there. It certainly was the easiest to install.
Okay, I get results when using the built in snmp tests...but whenver I use a test I added myself...I get no results. Just empty graphs.
As an example, I created an snmp test based on the Interfaces - Bytes In test. It's basically just the same test except I changed it to walk instead get and oid changed from ifInOctets.%ifIndex% to ifInOctets
I did this assuming I would get the results of all interfaces instead of just the one interface I would get with ifIndex. Instead, I get no results at all.
If I do an snmpwalk from my command line on the oid ifInOctets ...it works showing me all the interfaces.
I'm really stumped by this...I wish I could see the actuall command netmrg is using in the background to poll the device, maybe I could troubleshoot it from that way.
The "walk" type of SNMP test in NetMRG returns the Nth item's value encountered during the walk. It can't be used to get a total of all of the counters that were walked. It was intended for use with Windows processor indecies and the like, where they create a very large number for the index, but we only care about the Nth processor, etc.
Well, that explains alot. It's okay, I got it graphing correct numbers now. Since walk won't work for me, I just decided to monitor the gige ports on the supervisor and that's working great.
The return values I was getting were negative so I just made a change to the snmp tests to do ifHCOutOctets instead of ifOutOctets
Now the values show correct.
Can someone walk me through setting this up so I can get accurate numbers on my high speed interface both on 7200's and a 6509
Here's a rough overview
1) Goto the SNMP Tests tab; you'll need to add two new SNMP tests similar to the ifInOctets and ifOutOctets tests; instead use ifHCInOctets and ifHCOutOctets.
2) Goto the template group and the template linux box. Click on the 'Network Interface' and add two new monitors. They will be SNMP tests using the new tests you just defined.
3) Goto the template graph page and duplicate the LAN Traffic graph. Then edit this to change the name to something like 'High Data Rate Traffic Graph'.
Click on the graph to view the graph members. Edit each of these and select the monitors you just created in step 2.
At this point you can go view the interface cache for your Cisco device and use this template graph instead of the 'Linux - LAN Traffic' graph that is normally used.