Interface Traffic Graph not created for CentOS - Others OK

[jlar310]

We have been using Cacti for some time now with Windows, SUSE and Solaris hosts and everything has worked well. We've started working with CentOS now, but we can't get the network interface graphs to work for CentOS hosts. All of the other graphs work fine (CPU/Load/Disk/Mem). We are using the latest version of Cacti and net-snmp. We can't find any error messages anywhere. We just get a missing image for the interface graph.

Anybody have any ideas?

Thanks,

Jeff

[TheWitness]

You have to perform a verbose query and post your results. Also, walk the mib. I am sure it's something simple, like the snmp.conf file.

TheWitness

[Mark Susol]

I am sure it's something simple, like the snmp.conf file.

..or the cron task that uses "wwwrun php " instead of "wwwrun php5" which I discovered in SuSe10.1 as a problem and why the poller was never called. when I finally ran the line from command line I got the bash error about unknown command "php"

[jlar310]

You have to perform a verbose query and post your results. Also, walk the mib. I am sure it's something simple, like the snmp.conf file.

TheWitness

I don't know much about SNMP. I'm pretty much dependent on the plug + play functionality of Cacti. Can you give me a little something more to go on?

[jlar310]

I am sure it's something simple, like the snmp.conf file.

..or the cron task that uses "wwwrun php " instead of "wwwrun php5" which I discovered in SuSe10.1 as a problem and why the poller was never called. when I finally ran the line from command line I got the bash error about unknown command "php"

The poller is running. All the other graphs update for this and other hosts. Interface traffic graphs work for other hosts, just not CentOS machines.

[zheka]

Hi jlar310!!!

I just came accross your post. You say that you successfully create graph for your CentOS box. My interest is CPU. I cannot get them graphed at all. I have CentOS 4.2 and it has net-snmp installed and all my graph are just "none". Any ideas about what I am missing?
Appreciate if you give me any hint.

Sencerely
Eugene

[jlar310]

Hi jlar310!!!

I just came accross your post. You say that you successfully create graph for your CentOS box. My interest is CPU. I cannot get them graphed at all. I have CentOS 4.2 and it has net-snmp installed and all my graph are just "none". Any ideas about what I am missing?
Appreciate if you give me any hint.

Sencerely
Eugene

CentOS comes with a default snmpd.conf file that isn't suitable for Cacti. Read the Cacti docs for configuring a basic snmpd.conf

[zheka]

Good idea! I tried to look through cacti manual and didn't find any reference to snmpd.conf
Do you have ready to go snmpd.conf file for CentOS.

[jlar310]

Good idea! I tried to look through cacti manual and didn't find any reference to snmpd.conf
Do you have ready to go snmpd.conf file for CentOS.

http://www.cacti.net/downloads/docs/htm ... ml#AEN1900

That's pretty much all I've got in my snmp.conf except that I've added the IP address of the Cacti server to the "rocommunity public" line to make it a tiny bit more secure. See the snmp.conf man page for more details. The snmpconf program is also useful if not terribly intuitive.

[jlar310]

You have to perform a verbose query and post your results. Also, walk the mib. I am sure it's something simple, like the snmp.conf file.

TheWitness

OK, so I still don't know anything about "verbose query" or "walk the mib" but maybe this will help. We enabled more verbose logging in Cacti and it seems that SNMP is returning traffic stats -- so where's the graph?

Here is a snippet from the log for a CentOS host. I cut the lines short for clarity.

Code: Select all

cpu_nice, oid: .1.3.6.1.4.1.2021.11.51.0, output: 57
cpu_system, oid: .1.3.6.1.4.1.2021.11.52.0, output: 13837010
cpu_user, oid: .1.3.6.1.4.1.2021.11.50.0, output: 1329454
load_1min, oid: .1.3.6.1.4.1.2021.10.1.3.1, output: 0.67
load_15min, oid: .1.3.6.1.4.1.2021.10.1.3.3, output: 0.89
load_5min, oid: .1.3.6.1.4.1.2021.10.1.3.2, output: 0.90
mem_buffers, oid: .1.3.6.1.4.1.2021.4.14.0, output: 65460
mem_cache, oid: .1.3.6.1.4.1.2021.4.15.0, output: 3581312
mem_free, oid: .1.3.6.1.4.1.2021.4.6.0, output: 17644
traffic_in, oid: .1.3.6.1.2.1.2.2.1.10.2, output: 2341016426
traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.2, output: 572989108
traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.3, output: 1869116975
traffic_in, oid: .1.3.6.1.2.1.2.2.1.10.3, output: 0
hdd_free, oid: .1.3.6.1.4.1.2021.9.1.7.1, output: 16348372
hdd_used, oid: .1.3.6.1.4.1.2021.9.1.8.1, output: 12766420

[zheka]

Let me ask you, jlar310. are we talking about snmp.conf or snmpd.conf?
There is no way to configure RO or RW community string. I tried to run snmpconf answered all questions regarding snmp deamon and configuration parameters. I still can't get this OID that you mentioned in you post.

[jlar310]

Just as a sanity check, I created a completely new install of Cacti on a clean install of CentOS running under VMWare. All the client graphs including the interface traffic work just fine on the new install. Besides the Cacti host OS (SuSE vs. CentOS), the only real difference was the rrdtool version which was newer on the CentOS install. Maybe that's all it was...