cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 9

Any one having trouble with monitoring RHEL7 machines?

We are currently moving our policies to use RHEL 7 machines when ever there is a new machine added to our environment, but we started to notice that Solarwinds is having trouble with them.  At first we thought it was because we were using a minimal install of RHEL 7 that was the culprit, but we then made a newer template for our server that used a full install.  When we try to add the server to Solarwinds, with snmpv2, we get a test successful and we move to the next page where it looks for the interfaces and it just spins there.  It stays there for ~20mins and then it says the server can not be reached or is down.  We have made sure that internal firewall permissions allows snmp communication between our servers and the server and poller can communicate over ssh and icmp.  Any ideas for you would be greatly appreciated.

Thanks in advance.

0 Kudos
6 Replies
Level 13

To verify snmp communication from orion poller machine to target machine you can use snmp walk tool that is shipped together with orion (c:\Program Files (x86)\SolarWinds\Orion\SnmpWalk.exe). Besides basic connectivity it can also help you to determine whether there is some restriction on particular OID subsets for given communitystring or snmp user.

0 Kudos

Petr Vilem - I have verified that snmp communication from our pollers is possible, but I have noticed that with the snmpwalk tool, it works 1 in every 20 times.  This dosen't happen with our other RHEL machines (5 and 6).  But when it does have a success run I get the print our of the OID's and it is missing the ones I am looking for.  I am trying to get Solarwinds to report the CPU, Memory and Mount points.  I am not sure if this is because there is missing data in a OID database or snmp times out before we get far enough down the OID list.

0 Kudos
Level 13

The default snmpd.conf likely contains a number of "extra" items meant to add security around SNMP.  These items will look something like this:

com2sec      notConfigUser     default     public

...

group     notConfigGroup     v1     notConfigUser

group     notConfigGroup     v2c     notConfigUser

...

view    systemview     included     system

...

access     notConfigGroup ""     any     noauth     exact     systemview     none none

The easiest method to get around all this stuff is to move the existing snmpd.conf file to a new location, then create a new file from scratch and only put in the bare minimum:

rocommunity     <community>     <ipaddress>

sysLocation     MyLocation

sysContact     MyContact

So for example,

rocommunity     kragle     10.0.0.100

sysLocation     Bricksburg

sysContact     president.business@octancorp.com

You could also restrict access to parts of the SNMP tree by adding an OID behind the rocommunity line.  This restricts SNMP access to OIDs at the specified OID and below.  In my experience though, this just adds confusion and you end up not being able to get what you want.  Some high security environments might require this type of granular control though.

0 Kudos

I attempted this same configuration.  Still having issues with Soalrwinds returing a list of resources.  When I run snmpwalk from Solarwinds server to query the client server, I get a long list of attributes, but it works.  When I attempt to list resources on same client server within Solarwinds, nothing comes back - the "wait wheel" just spins.  I've gone through firewall, etc.  Firewall is turned off.  Using simple remote snmp tools to query works - but Solarwinds listing resources does not.  Anyone have any ideas?

0 Kudos

Branfarm - Thanks for this but we have a custom snmpd.conf file managed by puppet that does restrict the snmp access to only our pollers using the IP and community string and in some cases a snmp user with the snmpv3.  Fortunately we are able to secure our internal environment so that we do not limit access to the specific IP's to a limited range of OID's that it can receive from the server.

0 Kudos

Sounds good -- the reason I mentioned it is that we've had a recent issue that sounded very similar to what you're experiencing.  We started pushing out configs using Ansible, and someone added the OID restriction to the end of the community line and then was frustrated when nothing beyond the basic host information was available.

To rule out local snmp issues you might also try using snmpwalk from the machine itself and make sure you get all the data back you would expect.

0 Kudos