7 Replies Latest reply on Apr 19, 2012 9:35 AM by Karlo.Zatylny

    Configuring Dependencies

    karltbraun

      I understand how dependencies work and all, but I'm curious how most people go about setting up their groups.

       

      We have a main data center, and a bunch of sites connected via an MPLS network.  We have a single poller at the data center.

      My first impulse was to create a group for each site.  But when setting up the alerting dependencies, I ran into a problem.  If I lose contact with the site WAN router I don't want to get alerts from all the devices at the site.  I thought it would be easy to configure all devices in the site group as dependent on the site router.  But the site router is also in the group, and the dependency configuration won't let you make a group of objects dependent on a member of that group.

      So does that mean I have to further divide my site into groups (WAN Router, Core Switch, Downstream devices) and assign dependencies that way?  Seems like a lot of configuration for just the benefit of the dependencies.

      But I'm coming from a What's Up Gold background, and my perspective may be skewed by WUG's this-node-depends-on-these-nodes type of configuration (which is also tedious).

        • Re: Configuring Dependencies

          Hi Karl--

          Have your read this Technical Reference. It is helpful.

          M

            • Re: Configuring Dependencies
              karltbraun

              You know, I forgot I had that passed on to me already.  A cursory glance indicates it's going to do the trick.  I'll reply further if I need additional clarification.

               

              Thanks,

                • Re: Configuring Dependencies
                  karltbraun

                  After digging into this and discussing my needs with the support staff, it was determined that Dependencies were not quite ready for prime time, and we've implemented a rather crude method of dependencies by writting suppression rules.  This is, of course, cumbersome and time consuming.  At the moment, I only have this faux-dependencies set-up for devices at a particular site, with suppression if the site router is down.  I need more granularity than this now, and am wondering what the current status of dependencies is. 

                   

                  Have new write-ups been made since 2010?  If not, do the old write-ups (using groups) apply and do they work now?

                    • Re: Configuring Dependencies
                      karltbraun

                      *bump*

                        • Re: Configuring Dependencies
                          netlogix

                          I am using groups to do the dependencies and it works quite well.  I have dynamic site groups setup that use IP begins with x.x.x and I set that group to be dependent on the router which uses the loopback address, (can't use LAN address because it cant be dependent on itself.)  repeat for 50 sites - not much fun repeating but now all nodes I add are automatically setup with dependencies based on it's IP.

                          1 of 1 people found this helpful
                            • Re: Configuring Dependencies
                              karltbraun

                              Thanks.  I'll start looking into this.  I'm hoping for an "official" answer from SolarWinds as they were the ones that told me (some time ago) that it wasn't "ready for prime time".

                                • Re: Configuring Dependencies
                                  Karlo.Zatylny

                                  Hi Karl,

                                   

                                  I disagree that dependencies are not ready for prime time.  We have many large customers (thousands of nodes) using dependencies effectively.  While I agree that it is time consuming to create the dependencies for a large environment, that a well structured environment is fairly easy to create dependencies for.

                                   

                                  The write ups from 2010 still apply for how to set up dependencies and groups, as do many of my thwack posts about them.  There is often a separation on groups created for dependencies and groups created for general status roll up.  When we were developing the dependencies feature, we knew that people would want a way of grouping items together for defining dependencies.  At the same time, we were developing the Groups feature for status roll up.  It was efficient for us use the same grouping mechanism to achieve both of these goals.  Hence, you may have one set of groups that is for easy of use for status roll up - e.g. "Austin Lab" - but then also have different groups for dependencies - e.g. "Austin Lab Servers" that can be a child of your main router or switch.

                                   

                                  Using custom properties or naming schemes can make your group creation easier for more dynamic environments.  I have been on GoToMeetings with customers and seen a router go down and have hundreds of child nodes go unreachable.

                                   

                                  Please let me know if you have any questions.