neilmborilla replied on Sat, Mar 13 2010 11:38 PM

rated by 0 users

outsidesys:

• Node Polling:  What is used to poll a node?  I am assuming ICMP Ping, but I just want to make sure.
• Interface Polling:  What is used to poll an interface?  ICMP Ping, SNMP, or both?  If SNMP is involved, what specifically is polled?
• Volume Polling: I am assuming that SNMP is used to poll a volume, and I would like to know what specifically is polled.

Node and interface polling used both ICMP and SNMP. SNMP is involved in interface in getting information's such as IFname, description, MTU/speed, bandwidth, Rx and Tx, Packet loss, and many more

Volume Polling used SNMP to get Hard Disk Info, CD, Physical and Virtual Memory, and many more

many more-meaning it depends on your requirement and the MIBs and OIDs of that device such as disk queue, performance counter, etc

outsidesys:

 Node Status

There is a setting in the Web UI: “Unresponsive nodes are considered down after:  <N seconds>”.  Unresponsive in what way?  No response to ICMP ping?  No response to SNMP?  Both?

Both SNMP and ICMP

outsidesys replied on Mon, Mar 15 2010 10:46 AM

rated by 0 users

Thanks for the answers and clearing up my confusion on what is being polled for each type of resource. 

However, now I'm confused about what the pollers are doing when polling the status of an Interface or Volume via SNMP.

There are settings for "Polling Intervals" and "Polling Statistics Intervals".  When a poller is checking the status of an Interface or a Volume, is the poller checking a subset of SNMP data compared to Statistics polling?  If so, what specifically is being polled?

neilmborilla replied on Mon, Mar 15 2010 11:06 AM

rated by 0 users

Status Polls provide UP/DOWN (ICMP)

Statistics Polls provide detailed information such as Memory Usage, CPU Lode, Etc (SNMP)

You may find this link very helpful

http://www.solarwinds.com/support/orion/docs/gaps/gaps.htm

pacetti replied on Mon, Mar 15 2010 11:14 AM

rated by 0 users

outsidesys,

Node, Interface and Volume Polling Intervals apply to ICMP polls "to determine status and response time".

Node, Interface and Volume Statistics Polling Intervals apply to SNMP polls to determine "performance statistics".

For more information, see "Orion Polling Settings" in the SolarWinds Orion NPM Administrator Guide.

I'll make sure the SNMP/ICMP distinction is clarified in the next document release.

Thanks,

outsidesys replied on Mon, Mar 15 2010 11:52 AM

rated by 0 users

@neilmborilla,

Thank you for the link. Here's what I found from that documentation:

"Status polls provide the up-down status of an element. Statistics polls return detailed information, for example, memory usage, CPU load, and core temperature.

"Orion checks up-down status every 120 seconds. For nodes, this information is collected by ICMP ping. For interfaces and volumes, this information is collected through an SNMP request."

So that clears up how the status polling is working for nodes, interfaces, and volumes.

However, I would still like to know what the poller is requesting via SNMP to determine the status of an interface and a volume.  Can anyone give me OIDs? 

Having this information will go a long way when creating alerts.  For example, when I create an alert that has a trigger condition for volume status, I am given a choice of Up, Down, Warning, and Unknown.  Besides the Up condition, what specifically determines each of the other conditions?

Thanks again for everyone's input.

neilmborilla replied on Fri, Mar 19 2010 3:27 AM

rated by 0 users

outsidesys:

Can anyone give me OIDs? 

Interface = ifAdminStatus = 1.3.6.1.2.1.2.2.1.7

Volume = hrDeviceStatus.1 - n = 1.3.6.1.2.1.25.3.2.1.1 - n

note: - n = indicates the number of volume you have

outsidesys:

However, I would still like to know what the poller is requesting via SNMP to determine the status of an interface and a volume. 

Good question you have here, this is something SW staff can share with and I would like to add the ff:

Ex. Polling interval = 5 minutes. I have node A to Z, How would I know if all nodes have been polled completely for the set period of polling interval?

Is it for the first 2 to 3 minutes, it polls A to N then the succeeding minutes is for P to Z. Or the SNMP request poll should be broadcasted accross all A to Z nodes then poller will only seat and wait for the responses from nodes then write it to DB

darryld replied on Fri, Mar 19 2010 5:08 AM

rated by 0 users

The polling order seems to be determined by the order the pollers are listed in the pollers table in the database. Note that there may be multiple pollers required to collect all the statistical information for a single device or interface.

The Polls Per second Tuning setting controls how many of these polls are issued each second and therefore how quickly a full cycle of polling completes (assuming that all nodes/interfaces are set with the same polling interval)

Also if an element fails to respond to a status check, the element is retested every 10 seconds for (by default) 2 minutes