Sub-device parameters empty for ethernet
Everything is working, I am getting graphs, but not for eth0. When I re-cache the interface, I get the proper information for eth0 (IP and mac) but it tells me No such object available on this agent. I am assuming it means ifInOctets.%ifIndex% (which works from the command line using snmpwalk). If I go into the sub-menu for configured and dynamic parameters, I don't get any information regarding eth0 at all. Or is it using the wrong ifIndex number?
Did you graph this item using the 'monitor/graph' button under 'view interface cache'?
If not, edit the subdevice to make it an 'interface' and add a parameter using 'ifName' for the name, and 'eth0' for the value. After configuring these, the dynamic parameters should have values after the next gatherer run and the graphs should be created.
Adding ifName (value eth0) didn't fix the problem. But if I put ifIndex (with the proper index number) it works. Now for the weird part 1 server has 1 nic, and it works fine with the above settings. Another server with 2 nics, only 1 works, but they both have the same settings. And another server with 2 nics, with the same settings doesn't work at all. Actually a correction.....They all work for a lilttle while, then some stop loggin....with no error messages. And now after checking...just to make sure that they are not making a liar out of me...after not working for a pause, they all seeem to be graphing again! There is actually a break (white space) in the graph. I am going to wait awhile, just to see if they continue graphing.....
How much bandwidth are you pushing through these interfaces? I think our default templates might have a max of 100Mbps (12MB/s).
You also might have a flakey snmpd; we've seen that before. If you run on a binary package system, you may want to check for updates to net-snmp.
I did finally receive an error message
/usr/local/bin/netmrg_cron.sh line 24 /usr/local/bin/netmrg-gatherer syntax error operand expected (error token is "/usr/local/bin/netmrg-gatherer")
but haven't seen it since. This was about the time that all of the graphing was working, and then some died. I am also getting an error in /var/log/messages from snmpd which states No swap line in /proc/stat.
As for the bandwidth, I am not getting anywhere near 100mbps. I am testing the stress load on the one server, but nowhere near anything that could choke it. The version of net-snmp I am using is 5.1-7 on 2 machines ( it works fine for eth0 on the single-honed machine, and only eth1 works on a dual-honed machine) version 5.0.8 on a dual-honed (neither works) and 5.1.2-6mdk on my workstation, and eth0 doesn't work. It seems that when I manually run netgatherer, the information is logged and graphed.....but the cron job seems to stop without any errors...but not for all things. On the computer that eth0 works, so does CPU utilization and load average.
Sounds like you might be having a problem with NetMRG dying. What can you tell us about the server NetMRG is running on (architecture, OS, distro, etc)?
Do you feel comfortable getting a core file and doing a backtrace when NetMRG segfaults again?
After doing some digging, it appears that netmrg is not dying, the info is just not being graphed. I manually get the snmpwalk values to make sure that snmp is working, I check the graphs, and the values are being found, but still no graphs. So I checked the directory where the graphs are being stored......and most of them are owned by root. I change this, and now more things are being graphed. As time goes on, I will post if some things are not working.
Thanks for your help, guys!