AIX Monitoring

Hello,

I was wondering if any members of the community have experience with setting up AIX monitoring within LogicMonitor?

We have attempted to set it up but are having issues. We are getting basic stats, Host status, Ping and port 25 response time, nothing else.
I have checked the aixmibd log file and I am seeing this error – “Connect to SNMP agent on the host loopback failed, will keep trying"

I have run !snmpDiagnose from the debugger and get the attached. As far as I am aware there is no firewall blocking the SNMP communication, I have confirmed the service is running in AIX and have confirmed that the community string is correct in both LogicMonitor and AIX. As such I have a few questions -

1/ Has anyone setup AIX monitoring and experienced this issue?
2/ Is there a limitation on the characters, from an AIX perspective that can be included in a community string? H0rs3Sh0e - this isn't our community string but is an example of the character usage.

Any help is greatly appreciated.

Thanks,

ASCoates

Posted 4 years ago·Last reply 4 years ago

12 comments

LM User

·4 years ago

Are your SNMP credentials inherited/set the same way for non-working as for working servers?

Learner

·4 years ago

Yes , All below SNMP permissions are same in working & non working servers , now what should be the next step i am not getting ,

snmp.security
snmp.auth
snmp.authToken
snmp.privToken
snmp.priv

In working server snmpdiagnose gives result as OK & with nonworking Servers gives result as can't receive snmp reponse in time .

Learner

·4 years ago

Hi Ascoates & Stuart ,

i am also facing same issue with some of the AIX hosts where i am using snmp version as v3 , while running snmpdiagnose from collector debug window in step 1 detect system properties i get snmpget as timeout , snmpwalk as timeout & snmpbulkwalk as time out.

execution time - 45804 ms

result - can't receive snmp reponse in time .

Any suggestion will be highly appreciated , some AIX servers are working and some not .

checked with network Team & AIX team they are saying all good from their side then where is the problem ?

LM User

·4 years ago

Most likely the target system has view restrictions on that community. Instead of allowing that community to view the required OIDs, that particular community is locked down to only certain oids. Not being an AIX guy myself, i couldn't tell you how that is done in AIX. But somewhere in the SNMP config, there's likely a restriction configured, locking it down to only certain OIDs.

ASCoates

OP4 years ago

Thanks again for the reply Stuart.

Did the as you advised and this time get the following -

So it looks like the community string is working, but for some reason the SNMP data isn't being reported back?
Could this be because something is blocking SNMP access?

LM User

·4 years ago

Ok, when using the collector debug console, specify the device by the entire system.hostname or the entire system.displayname (system.hostname is usually easier). Also, the system.hostname should be resolvable because that's the address to which communications are directed. I don't think using the IP address will search through system.ips.

ASCoates

OP4 years ago

39 minutes ago, Stuart Weenig said:

Look at the line right above step 0. Last part of that line.

If the community string has been set as a property on that device, but the debugger isn't seeing it, there might be several causes:

1. You're debugging on a Collector that's not the same as the Collector monitoring that device. Collectors only have access to the properties of devices assigned to them.

2. You're using 172.25.10.4, which might not be the displayName nor address of that device. Even if it's an IP address on that device, if it's not the displayName nor the address, the Collector won't match it up with a currently monitored device and will think you are trying to diagnose something that isn't monitored in LM. In that case, it doesn't think there are any properties to look up, so it uses the default "public"

As a side note, that many "collector has fetched the task" messages sticks out to me. You might look to see that your collector is performing well.

I'm blind and completely missed the "Public" line.

I have changed to the collector allocated to the server in question and I got the same result. I have also checked the device info in LogicMonitor and found it contains the correct information -

ASCoates

OP4 years ago

I went back and did another !snmpdiagnose, this time with the snmp version and community passed in the command and I got what appears to be the expected result -

Am I correct in my thinking that this means the SNMP Community is correct in AIX and for some reason, on the device in LogicMonitor the community is not being passed?

LM User

·4 years ago

3 minutes ago, ASCoates said:

Question: How can yo tell from the diagnose that it is trying to use "Public" as the community string?

Look at the line right above step 0. Last part of that line.

If the community string has been set as a property on that device, but the debugger isn't seeing it, there might be several causes:

1. You're debugging on a Collector that's not the same as the Collector monitoring that device. Collectors only have access to the properties of devices assigned to them.

2. You're using 172.25.10.4, which might not be the displayName nor address of that device. Even if it's an IP address on that device, if it's not the displayName nor the address, the Collector won't match it up with a currently monitored device and will think you are trying to diagnose something that isn't monitored in LM. In that case, it doesn't think there are any properties to look up, so it uses the default "public"

As a side note, that many "collector has fetched the task" messages sticks out to me. You might look to see that your collector is performing well.

ASCoates

OP4 years ago

22 hours ago, Stuart Weenig said:

My guess is this is not related as it seems like a message indicating that the AIX system couldn't connect to itself over SNMP.

The fact that the AIX system couldn't connect to itself over SNMP may indicate that you don't have the SNMP daemon setup to listen for or allow connections over SNMP. This is confirmed by your !snmpdiagnose output.

Your !snmpdiagnose indicates that LM is trying to use 'public' as the community string to communicate with your AIX server. Hopefully you're not actually using "public" as your community string. Take a look here to see how to give LM the right credentials.

Thanks for the reply Stuart.

Your last line does make me remember another article that I had forgotten about which mentioned running the services but with a command that attaches the community string to the services. I will give this a try and see if it works.
I know for a fact that the community string is correct and they have been set as a custom property on the device (the same way we have for other devices)

Question: How can yo tell from the diagnose that it is trying to use "Public" as the community string?

Thanks,

LM User

·4 years ago

4 hours ago, ASCoates said:

I have checked the aixmibd log file and I am seeing this error – “Connect to SNMP agent on the host loopback failed, will keep trying"

My guess is this is not related as it seems like a message indicating that the AIX system couldn't connect to itself over SNMP.

The fact that the AIX system couldn't connect to itself over SNMP may indicate that you don't have the SNMP daemon setup to listen for or allow connections over SNMP. This is confirmed by your !snmpdiagnose output.

Your !snmpdiagnose indicates that LM is trying to use 'public' as the community string to communicate with your AIX server. Hopefully you're not actually using "public" as your community string. Take a look here to see how to give LM the right credentials.