3 Replies Latest reply on Dec 4, 2014 2:14 PM by ma_pmchenry





      We've created alot of alerts and they are helpful and work just fine but, I would like to clean up the alerts so we don't get hit with 20 emails when we know the site is down. This seems easy enough, but we have backup connections to our remote sites so, I can't just create a group and make all of the devices defendent upon the WAN router.... for example.


      Because we have backup VPNs to all of the remote sites as well I would have to make a group that made all devices dependent upon the WAN router and the site's VPN(Internet) connection.So, if the WAN router and the VPN were down, then we could stop alerts from the rest of the devices at the site.


      Also, I noticed that by default, interfaces that are part of a failed device send alerts when the device goes down. I would like to stop these alerts as well.


      Could someone please point me in the right direction to start configuring these dependencies - a good doc?


      Thank you

        • Re: Dependencies

          Pat McHenry


          I have found it is a timing issue, if the system finds the child object down prior to the parent you will get the alert. This can happen if the polling time is faster for interfaces than Nodes. I usually will stack my polling from most critical to the least, which works well when the alert engine looks for Down Nodes, Interfaces, etc.. Here is an example on Router to Switch to Server at a remote site.


          Status Polling Times

          1 Minute Router

          1.5 Minute for Switch

          2 Minute (Default) Servers


          Best Regards,

          Derik Pfeffer

          Loop1 Systems: SolarWinds Training and Professional Services


          • Re: Dependencies

            Hi Pat,


            we had the same problem as you, one thing is very important.


            ----- Never create more than 600 dependencies, this will greatly slow down your Application ( edit a node within 50 sec and more )! ------


            We had to create "device" Groups and then build the dependency for each group - not for each Device!



            A Group for Routers (dynamic filter or manually assigned) and a Group for other Devices (dynamic filter or manually assigned)

            Then build Dependency

            Group Devices depend an Group Router


            (Group Routers including the Backup Connection, so when both - Router and Backup Connection - are Down the Group status changes to down and will trigger the dependency for the devices!)


            Now when the Router stops responding or WAN is down, the Devices will stop reporting and have the status "unreachable".

            They will report again when Routers are up again.





              • Re: Dependencies

                Alois -


                Some of my dependencies I've created don't work.What I ended up doing was creating a parent/child relationship called "MPLS" between the WAN router and group called "Site Devices". So, the MPLS router is the parent and a group that contains the rest of the site devices, excluding the Internet/VPN backup router.


                Then I created a Parent/child relationship called Site "MPLS and Internet" the parent is a group called "MPLS and Internet" containing the MPLS router and an IPSLA pinging an Internet address from my Core switch at the site. The child is a group called "Site Devices" that includes all of the sites devices except for the MPLS router.


                The last Parent/Child relationship I created is called "Core Switch". The parent is the Core Switch and the child is a group called "Device Behind Core Switch"


                I haven't tested the first two dependencies, but I did test the third the other night when I upgraded a the sites' Core switch. It didn't work.I still got alerts from all of the devices behind the Core Switch.


                I also changed the polling of the Core Switch to 1 minute and left the devices behind the core switches default.


                In your last post are you suggesting that I must configure the Core switch as a group and use that in the dependency rather than the node?


                Thank you, Pat