I need help with Sun SNMP

[emota]

Hi guys,

I'm having a problem monitoring some servers that run Sun. My problem is that I have to monitor the CPU Load, Memory and Disk usage on the servers. I have to do it using SNMP, currently the servers have the default Sun SNMP agent running and I don't know and haven't found the OID's for these things. I'm going to monitor them using RRDTool and cacti. I'm monitoring other servers that run UCD-SNMP and I didn't have any difficulty. My problem is with Sun SNMP (and I'm not allowed to change it). Does anyone know the OID's for memory, CPU and disk usage of Sun SNMP or do you know another way I could monitor these things not changing the SNMP agent used currently? I'm new on Solaris/Linux so I would appreciate any help you could give me.

Thanks,

Ede

[sayasif]

Hi,

You will get it here ..

http://www.mibcentral.com/cgi-bin/vendo ... n&id=22858


or you can try exploring the mib2 tree .. if you gotta a good MIB browser then things gets easy to find ..

try using this
http://www.wtcs.org/snmp4tpc/getif.htm

hope that helps

ACE

[emota]

Thanks for your help.... I found a mib file in the page you indicated that has OID's for memory, disk and CPU usage, the file is kernel-reader-mib.mib, the servers that I have to monitor don't have that mib file, so I guess I have to add it to the snmp agents. Do you know how I can do that?

Ede

[sayasif]

Hi,

I'm glad that it helped you

not very sure how to install SUN specific Mibs. but guess it would be easy if you install the SUN native tool for monitoring on the server(you gotta find this on the sun site) its something what CIM(compaq insight manager) agent does for compaq servers.

Coming back to the MIBS for CPU, memory and DISK for Sun systems.. I got the following for cobalt server and it should be the same for your sun server also.

CPU Idle time
.1.3.6.1.2.1.25.3.3.1.2.1 (one minute average)

Memory total
.1.3.6.1.2.1.25.2.3.1.5.1
Memory Free
.1.3.6.1.2.1.25.2.3.1.6.1
swap total
.1.3.6.1.2.1.25.2.3.1.5.2
swap free
.1.3.6.1.2.1.25.2.3.1.6.2

disk space
.1.3.6.1.2.1.25.2.3.1.5.xx where xx is teh disk instance
disk free
.1.3.6.1.2.1.25.2.3.1.6.xx where xx is the disk instance


I would suggest you to write cdef to get the percentage values for above so that you can show it on a scale of 100 for example to graph disk usage
if diskspace1=a
,diskspace2=b,diskspace3=c and diskspace1used=d,diskspace2used=e,and diskspace3u=f

then you can write 3 cdef to get individual disk percentage against the total disk capacity

for example for disk 1 used percentage

d,a,b,+,c+,/,100,*

similarly you should do for the other two disk.

Now you can put disk 1 as area and disk2 and disk3 as stacks which makes a pretty graph. Best part is that you can reuse the cdef provided your data sources are in the correct alpabetic order and you have put a custom entry with alpahets in the CDEF

Let me know if you need more help on this.

best of luck

ACE

[chrism]

Hi,
We've just crossed this bridge. We have a number of E10K's, E450's etc all running Solaris 8. For some reason they seem to have the standard Sun snmp mib (look for it on www.mibcentral.com) installed, and not the sun management centre mib. The SunMC mib gives useful info such as Load Average, processor interrupts etc - although you can work out headline CPU busy from the regular mib.

Set up the following as SNMP data sources using Derive or counter.
.enterprises.42.3.13.3.0 - System time
.enterprises.42.3.13.3.0 - User time
.enterprises.42.3.13.3.0 - Nice Time

Create a CDEF to normalise the CPU counts by the number of processors on your box. (i.e. for a CDEF on a 64 processor starfire, we have 'Data,64,/').

Create a chart with (for example) system as the bottom area, and the other two stacked above.

We're currently monitoring around 20 servers using this technique, and it works well. I'm not sure if we can use the Disk figures, as we use Veritas VM to manage storage. More experimentation needed.

Chris M