4 Replies Latest reply on Nov 20, 2008 4:53 AM by pdjhh

    Polling engine NAT'd

    pdjhh

      Hi,


      We built a polling engine that is in a secure environment. It is using a database on a central NPM but there is a nating firewall between the two. The remote polling engine shows up when I launch the system manager on the central NPM but it shows it's real address and not it's NAT address. I can go into the DB and change the address to the Nat address but as soon as you start the services up again on the polling engine it changes it back to the real address so no communications can run from the central box to the polling engine. I have changed the local hosts file to reflect the nat address but exactly the same thing happens. Surely it is reasonable to have a poller in a secure nat'd environment? Any ideas on this?


       


      Thanks.

        • Re: Polling engine NAT'd
          denny.lecompte

           We definitely have users successfully running NPM through NAT.  We expect the polling engine to show the real IP address, not the NAT address, but it should work if the right ports are open. 

           What's communication is failing?  If you're trying to use System Manager remotely, that's not going to work, NAT or no NAT.  System Manager is not remote-able for a number of reasons.  If you want to manage remotely, most things can be done via the web console.  If it can't be done through the web, you have to log in to the local box.  In future releases, everything will be manageable through the web console.

            • Re: Polling engine NAT'd
              pdjhh

              That's good informatin thanks. Yes I guess we were trying to manage the remote poller using the sytem manager on the primary NPM but then it does give it as an option! So if I do all my configuration using the web tool then I get funny outcomes too. So I'm on the cental server and I add a node. The remote poller is in the drop down list so I select that poller for the device I want to add. It then offers to validate the snmp. So I click validate and it fails. I run the sniffer and can see that it's trying to validate the snmp from the central server even though I've selected a remote poller! SNMP isn't permitted to that device from the central server that's why we've got remote pollers. No one from Solarwinds support ASIAPAC seems to have any idea how to configure this, as you agree, reasonable configuration. If you're able to assist in any way that would be great. We're stuffed otherwise I think.


               


              Cheers.

                • Re: Polling engine NAT'd
                  pdjhh

                  I just got this from Asiapac support, are they correct?


                  I have got confirmation from my bosses that accessing the remote pollers from your main poller is not going to work.
                   
                  The pollers themseleves communicated via DCOM ports which run on random ports. So if you are running this behind the firewall is not possible to ge this to work
                   
                  Hope the above clarifys

                • Re: Polling engine NAT'd
                  pdjhh
                  Hi,

                  In reponse to my recent post talking about managing my nat'd polling engines from the central NPM through system manager, Solarwinds have told me this won't work and to forget about system manager. They say the next patch will allow me to manage my remote pollers through the web interface so I'll just wait for that. Another outstanding one I have though if you could please have a think, My remote polling engines are sending basic data back to the central NPM and all is well. However when I define a new poller through the universal device poller on the remote polling engine, after I click the finish button to define it I always get 'could not reach jobscheduler service after 3 attempts'. I don't know if it's trying to talk to it's own job scheduler or the remote one but I get the message anyway. I can see the newly defined poller back on the NPM and can assign it succesfully but it never gets any data ( a test from the remote poller runs succesfully so the oid is correct ). I've tried all the solutions I can find on that error and nothing's fixed it. I've logged a case with Solarwinds Asia and they can't work it out. The whole solution is useless to me without custom pollers!!