Using a network discovery to add volumes for monitoring

We have hundreds of Agent-polled nodes and I want to add Swap space monitoring for them.
This is important: nodes are already Agent-polled.
Unfortunately, "List Resource" button is hidden if I check several nodes:

This is example of bad UX, where I can't use ordinary feauture without any reason.


Likely, you will suggest me to use Network Discovery, but it doesn't work well.

My step-by-step course of action:

1. Settings > Network Discovery

And the first question: how can I add nodes here automaticly (for example based of NodeRole or NodeOwner custom property)?
I have no list if IP's for my dozens nodes.
Manual filling IP addresses is nonsense. 

2. Next, Next, Next

3. "Manually set up monitoring" (because I don't want to add volumes other than Swap space): 

4. Yes, run:

5. Waiting some time

6. Results (5/5 nodes found - I wouldn't have the patience for 250 IP's)

 

And the second question: Why are they "net-snmp - Linux"? They are Agent-polled nodes.

7. I don't need any network interfaces, so ucheck all:

8. Volumes. Let's check only "RAM", because Swap space is in this category.

  

9. Next, Next, Next

10. Results:

Skipped, not processed. That's pretty suspicious, hm?

11. Finish

12. Let's check these 5 nodes, shall we?

First one:

No tick!


Second:

No tick!


Third:

No tick!

...


Fourth:

No tick!


Fifth:

No tick!


So question is the same: how to add volumes for several (dozens/hundreds) nodes simultaneously?

  • So probably the only way to do that would be through the Orion API and I'm not smart enough to guide you through that one. The reason the nodes are skipped is that as stated they are already in the Orion DB.

  • Hello, 

    First a sanity check, your process is fine as I followed it myself and didn't have any problems adding volumes to existing nodes:




    1. (Note the lack of the 'Import Status: Not Processed' message - here is the key)



    That leads me closer to believing there is some other system that is interfering, and my bet would be on the 'Duplicate Detector' feature, which you can try disabling from here: https://support.solarwinds.com/SuccessCenter/s/article/Disable-Duplicate-Detector-for-Discovery-Engine?language=en_US

    If you are still having problems, I'd recommend raising this to support as they will be able to go through the discovery logs to see exactly what is going on. 

    Kind regards,

    Marlie Fancourt | SolarWinds Pre-Sales Manager

    Prosperon Networks | SolarWinds Partner since 2006

    If this helps answer your question please mark my answer as confirmed to help other users, thank you!

  • Hello, thank you for answer.

    didn't have any problems adding volumes to existing nodes

    Can't really see a tick near a new volume on your screenshots. 
    Yes - I have all volumes in the Resource list (including desired "Swap space" volume) - but without a tick this is useless Slight smile

    Case is already created (00928803) but to be honest I have no illusions about the quality of technical support service Disappointed
    All complex issues have been resolved on our own or with Thwack or haven't been resolved at all.

  • I can indeed confirm that it was all ticked successfully in the 'List Resources' section too :) :

    It IS possible for a volume to be in the database but not being ticked in the 'List Resources' section. You can run 2 queries to see what is going on using the Database Manager: support.solarwinds.com/.../How-to-Run-a-SQL-Query-in-Orion-Database-Manager-on-Windows-Server-2008

    Query 1 will check to see which volumes are listed in the database under the specific node:

    SELECT v.volumeid, v.caption, v.volumetype FROM [dbo].[Volumes] v
    join [dbo].[nodes] n on v.nodeid = n.nodeid
    where n.caption = '$NODECAPTION' --update $NODECAPTION for your device

    Query 2 will show whether or not all the pollers for those volumes are currently running (AKA are ticked in 'List Resources') - for each volume there should be a 'V.Details...', 'V.Statistics...', and a 'V.Status...' poller listed here and with the 'Enabled' box ticked:

    SELECT TOP 1000 * FROM [dbo].[Pollers]
    where netobjecttype = 'V' and netobjectid in (SELECT v.volumeid FROM [dbo].[Volumes] v
    join [dbo].[nodes] n on v.nodeid = n.nodeid
    where n.caption = '$NODECAPTION') --update $NODECAPTION for your device

    Kind regards, 

    Marlie. 

  • Hi. Not sure this is very useful, but these are results:

    Swap space is presented with volumeid 2049:

    The same id is presented as poller:

  • So there's a bit of a learning curve when it comes to re-discovering existing agent monitored nodes and some notable points that help...

    • Don't add any addresses to the discovery (this is the biggest hangup / UI fail) - you MUST have an address. I put a single address thats blackholed on the subnet or you could do localhost.
    • Agent discoveries are tied to the polling engine the discovery profile is assigned ( 3 engines require 3 separate profiles to rediscover all agents )
    • See the Agents tab in the discovery to configure agent discoveries selection criteria
    • You must check the poll / scan for existing agent polled nodes on the agents tab when configuring your discovery to scan them (this will enable some dynamic selection ... keep in mind its specific to that engine and NOT looking at all nodes)

    Your nodes are showing as net-snmp because your scanning them remotely via snmp (by providing the ip addresses) so the linux agent monitored nodes likely are not restricting their snmp traffic to its own localhost which is usually suplemental of the agent itself (linux agent nodes allow for additional snmp creds to be supplied)

    If you disable that duplicate detector or also enable duplicate nodes you'll end up with just that... duplicate nodes which is likely not what you intended.. e.g. server-xyz monitored via agent and server-xyz monitored via snmp as well.

  • Thank you, Network Discovery has become clearer to me

  • btw today I had a webex with tech.support, they tried to perform a discovery to add 'swap space' volumes.
    Because of these actions I have about 200 FULL-checks hosts now:

    This is a bit weird (firstly because tech.support doesn't know what he's doing Open mouth and secondly because our poller is overloaded now).

    How to fix this for all 200 nodes?
    I'd like to monitor only certain resources, not all:

  • Hi there, 

    You can un-manage these by running these queries in order:

    PLEASE RUN THESE AT YOUR OWN RISK AFTER TESTING, NOT LIABLE FOR ANY PROBLEMS THAT OCCUR

    First, un-tick all of the pollers whose volumes don't match the name 'Swap Space':

    --Make a backup of your DB before running ANY UPDATE/DELETE commands
    
    UPDATE [dbo].[Pollers] p
    INNER JOIN [dbo].[Volumes] v on v.VolumeID = p.NetObjectID 
    SET p.Enabled = 'False'
    WHERE p.NetObjectType = 'V'
    AND v.VolumeDescription <> 'Swap Space'

    Second, remove all volume objects that are were unticked from the first query:

    --Make a backup of your DB before running ANY UPDATE/DELETE commands
    
    DELETE FROM [dbo].[Volumes] v
    Select * from volumes v
    INNER JOIN [dbo].[Pollers] p on v.VolumeID = p.NetObjectID 
    WHERE p.NetObjectType = 'V' and P.Enabled = 'False'

    Finally, remove the un-ticked pollers from the poller table (This is optional as I think the database maintenance would sort this out overnight):

    --Make a backup of your DB before running ANY UPDATE/DELETE commands
    
    DELETE FROM [dbo].[Pollers] p
    WHERE p.NetObjectType = 'V' and P.Enabled = 'False'

    Kind regards,

    Marlie Fancourt | SolarWinds Pre-Sales Manager

    Prosperon Networks | SolarWinds Partner since 2006

    If this helps answer your question please mark my answer as confirmed to help other users, thank you!

  • You need to be very careful with discoveries... running this (especially with auto-import) can cause horrible consequences for your Orion environment by adding interfaces, volumes, nodes, pollers and/or applications or more to your setup without you wanting it to. They need to be scoped appropriately and always testing by limiting your scope first.

    - that's spot on. I'd only add that there's more than one way to skin a well you know. :) I prefer the API if I can and fall back to SQL if there's no API means to do what I need to do.

    If interfaces were added you may need to do a similar query for those as well to remove.

    Here's a SWQL query (note SWQL... not SQL) that you could utilize to pull all the interface and volume netobjectid's of recently added ones.

    SELECT e.NetObjectID, e.Message, e.EngineID, e.EventTime, e.EventType, et.Name
    FROM Orion.Events AS e
    INNER JOIN Orion.EventTypes AS et ON et.EventType = e.EventType
    WHERE e.EventType IN ( 16, 202 )
    -- AND e.EventTime > '2021-11-01 00:00:00.000'
    ORDER BY e.EventTime DESC

    Hope you make it through this, there's light at the end of the tunnel we promise!

    cheers