2 Replies Latest reply on Nov 15, 2010 11:37 PM by ameruz

    Network Sonar behavior on repeated discoveries - RC4

    ameruz

      When I run Network Sonar for the first time on a subnet, it works great, I add all the nodes/elements and so on.

      Some of the nodes are Linux servers and each server has dozens of NFS shares mapped, which all show as volumes. We don't need to monitor any of them as we monitor EMC filer separately, we only want to monitor local disk partitions.

      I then need to manually select all the volumes, separately for each node (server) and delete them - VERY time consuming.

      Then, next scheduled discovery will run and it will discover all the NFS volumes again and If I accept that, they will be put back in the database.

      If, at the 2nd run of discovery, I marke the node which has NFS shares to be ignored, it won't be discovered again.

      A couple of issues with that: I need to know which nodes have NFS shares and which not, which is very difficult when you have 350 servers, and some only have a couple NFS shares or none and some have 26.

      Second, if I ignore the node, then nothing else on that node will be discovered again. Sometimes, we enable additional network interfaces on the servers and we won't see them in Orion.

      Keeping track of differences by hand defeats the purpose of running automated discovery.

      Better choice would be for NPM to allow option to ignore specific associated elements with the node (interface, volume etc.) and not the whole node only.

      Or it should recognize that elements were manually deleted after discovery has run and not add those same interfaces back.

      This may not be a new behavior to 10.1, but I never tried this until I installed 10.1 the other day.

      In the past, I never trusted the discovery, but I wanted to use auto discovery this time to test "Connect now" feature in network atlas ....

      Appreciate any feedback.