Converting from External polling method to ICMP polling method should add ICMP status poller

I ran into a weird issue where if you have an external node and it is converted to the ICMP polling method, the system doesn't add the ICMP poller in the background. The node status stays Unknown because it is never polled.

The "fix" is currently delete the node and re-add it back in as an ICMP polled node. This would be fine unless you have SAM monitors assigned to the node - these are then also deleted and you lose all the historical data. You could probably add the poller manually via the API or the DB but you shouldn't have to.

I reported it as a bug to support (Case # 00229327) but I'm not sure if there were some wires crossed but the response was basically "working as intended". I get that external nodes don't gather statistics, that's why I'm converting it to an ICMP node! If this is not supported, the option to convert an external node should be removed.

" this behavior is by design. External nodes are not gathering any statistics. To enforce the node to gather statistics, the External node needs to be removed and re-added." This has already been reported and answered on our Developer case: CORE-4584

Find more posts tagged with

external node

core

Status: None

Comments

designerfx

I remember this and I get the idea, and naming wise external nodes should be called unmonitored nodes. External to the monitoring system is implied, but that wording is strange.

People should be able to go from external -> monitored as it were, but it'd be more logical to say we're going from "unmonitored -> monitored".

HerrDoktor

shuth, while I find your Idea great and it seems to be a no-brainer, in the meantime there is something you can do about this on your own:

here's how to add a ICMP Poller via the SDK, you need to adjust your select statement to only cover the nodes you want.

this is just quick and dirty but should give you an idea how to work with the SDK

if (!(Get-PSSnapin | where {$_.Name -eq "SwisSnapin"})) {

Add-PSSnapin "SwisSnapin"

}

## Orion System ##

$hostname = "localhost"

$username = "admin"

$password = "admin"

$cred = New-Object -typename System.Management.Automation.PSCredential -argumentlist @($username,(ConvertTo-SecureString -String $password -AsPlainText -Force))

$target = Connect-Swis -Credential $cred -Hostname $hostname

$SystemName = ''

## END Orion System ##

$AllNodes = Get-SwisData $target "SELECT Uri FROM Orion.Nodes"

foreach($Node in $AllNodes){

$NodeProps = Get-SwisObject $target -Uri $Node

$Pollers = Get-SwisData $target "Select PollerType, NetObject, NetObjectID FROM Orion.Pollers WHERE NetObjectID=@ip" @{ip=$NodeProps.NodeID}

$PollerFound = $false

foreach($Poller in $Pollers){

if($Poller.PollerType -eq 'N.Status.ICMP.Native'){

$PollerFound = $true

}

if(!$PollerFound){

# register specific pollers for the node

$poller = @{

NetObject="N:"+$NodeProps["NodeID"];

NetObjectType="N";

NetObjectID=$NodeProps["NodeID"];

}

# Status

$poller["PollerType"]="N.Status.ICMP.Native";

$pollerUri = New-SwisObject $target -EntityType "Orion.Pollers" -Properties $poller

}

bobmarley

I recently ran into this as well. Thanks you HerrDoktor for providing a work around.

HerrDoktor

No problem. I had this issue when I forgot to add the ICMP Poller in another script... Customer was complaining that all Nodes were up in Orion even when they were clearly down... big mistake by myself, but I found it rather quickly and was able to write this short fix for it

shuth

Cheers HerrDoktor. I've used a similar script to fix this issue. This FR is mainly for those users unfamiliar with the SDK so your script helps provide a good workaround in the meantime.

FormerMember

I have exactly the same problem. I will look at using this fix. Thank you for the post

nitrus73

Same issue here. I initially added a number of Oracle databases into SAM and the documentation I was following recommended that I add the Clusters as External. Then I discovered that to get DPA to link to SAM the Clusters have to be polled as ICMP for the software to correlate the two together. Tried changing the polling engine and also the polling method to SNMP and back to ICMP to no avail and I really don't want to lose all the historical data in SAM and have to add all these databases again!

I am not familiar with using the SDK method mention above and would prefer not to use a method that is "not officially recognized".

This is something that really needs to be fixed!

nitrus73

Regarding changing the polling method from ICMP to SNMP and then back. I was wrong. It DID work and my clusters are now showing as up and I did not lose any of my data that SAM had collected.