5 Replies Latest reply on Oct 24, 2011 3:09 PM by jpl3

    UDT - testing issues and requests

    BryanBecker

      I'm kicking the tires on UDT and have a few issues I'm come across, maybe someone has posted this in another thread.  I also added some nice to haves in a future release.


      ·     Under Node Details the section/view for "Port Details device xyz" is show all ports, same for th node tree in the All UDT Node view.  I marked all the VLAN and a few other ports unmonitored in UDT but they are still show up in this view


      ·     Cosmetically that same view needs better column sizing.  Things do not line up orderly in this view.  As with NPM in general I wish SW would allow us to strip off the domain name of a device if we want to.  In our environment the domain name can be quite large


      ·     Under Device Tracker Port details the Port History section is not displaying the hostname of the device (just IP and MAC), yet in the Node details it is for the same port.


      ·     I find it redundant that you have to re-add devices to UDT that are already being monitored by NPM.  Like NCM, UDT should automatically pull devices (sync) from NPM so you don't have to do it twice.  All ports at this point should be monitored except thinks like Layer 2 VLANS, EO0/0 ports, port channel ports, etc.  These should be disabled by default.


      ·     When you hover over a port, a couple pieces or information show be in the pop-up window.  I would add current IP(s) and hostanme of the devices connected, and if it's a windows machine the last logged in person's account ID


      ·     I'm finding searching for mac address or IP of a device sometimes doesn't not show up in the look up results.  I can see both when looking at a port and neither active or historical searches show the port.  Device Tracker search results mostly does not work.  I can click on a device it knows its name and IP and in the search results it shows nothing


      ·     Hostnames are populated for windows devices, I'm assuming using dns lookups, but UDT will not display Cisco IP Phone DNS names.  We are in DNS so why isn't this being displayed?


      ·     Another feature request would be notification or something displayed on the port list view that the port has not had any devices connected to it after X period of time.  We patched all primary ports for every cube we have and out closet switches are almost out of space.  It would be nice to know this so we can trace this port to a wall jack and remove it to clean up the switch patching.  We have a switch that I know is patched with about 10 ports left physically open yet UDT says 40% of the ports are free.  I think a 30 day or something like that counter would be nice so we can clean up the ports.


      I look forward to the AD integration that will allow us to see the last logged in (or current) person is to a particular port.    If any of these issues is user error on my part please let me know.  Thank you.


      BB

        • Re: UDT - testing issues and requests
          mavturner

          Bryan,

          Thanks for the detailed response here.

          1. This is something we are aware of and are quickly working on two things to make this better. The first is being able to delete a port, the second is the provide filter options on the Port Details resource so you only see specific types of ports (active, monitored, specific ifTypes, etc.).
          2. This is a good request.
          3. I'll let someone from dev comment on this, they may want more information to see what is going on.
          4. The nodes are automatically monitored, but you need to tell us what ports you want to monitor. Most people are not monitoring all of their ports in NPM, so if we only picked up on those ports, we would not have full connection information. The wizard should be a fairly simply Next Next situation for customers who are already managing the nodes the want to monitor ports on. The nodes in UDT are exactly the same as the NPM nodes in this integrated case. What would you prefer, automatically add all ports for all currently managed nodes?
          5. Good request. 
          6. We have a hotfix that I will send you offline that should help with this information. I assume if you change the connection type from Direct to All, you will see the device it is connected to (as well as all other switches that are in the same layer 2 domain). This is something we are continuing to make improvements on and fixing environmental issues as they arise.
          7. We use the standard Windows name resolution process. If you do an nslookup form the server, that should be equivalent to what UDT is reporting. See this blog post for UDT Tips and Tricks.
          8. We have some reports that should help with this (no activity for X days). Modifying the other resources to show this information is interesting too.
          I look forward to your futher feedback!
          Mav

            • Re: UDT - testing issues and requests
              michiganlaz

              We just installed this today and while I did a happy dance, I was very frightened when I discovered that the guest account could run a discovery!

              I poked around and saw no way to limit accounts to 'read-only' in UDT- a very, very necessary feature. Can't have just anyone running a discovery, but we do want to let operators, the security team and other non-engineering folks search for devices. Maybe I'm missing it, please advise if I am!!

              Thanks. It really is a wonderful product and even in its first iteration it is incredible and fills a HUGE need here!

              Also- under discovery filtering, it would be helpful to automagically ignore vlan objects and the stack ports on 3750s. Perhaps there's a way to do that with regex but I'm not proficient in that, though I intend to mess around and see if that will do the trick for me.

              Thanks!

              Edit- I read your tips and tricks after the fact and see a way to limit which ifTypes to discover. I'll give that a go, but I hope long term there is a better way to select ifTypes via the discovery dialog. Thanks.

              • Re: UDT - testing issues and requests
                BryanBecker

                Mav,

                TY for the resposes.  I had some follow-ups to these :

                • #4 - Unless I'm missing it the nodes are not automatically monitored.  My test NPM server has about 100 devices on it, yet in UDT it shows only 6.  And those 6 I had to run the wizard to get them added.  I understand of the NPM monitoring versus UDT monitoring.  We only monitor network to network ports and some key server ports but UDT is monitoring them all.  I'm good with that.  I see the UDT part of the wizard and that's good too.  I'm just saying it might be best to keep NPM and UDT in sync for nodes and allow the user to pick and choose ports to monitor, etc.  That's for exisiting NPM servers where you later add UDT. 
                • #7 - This is a weird one.  I'm on our test NPM server and from cli I cannot do a nslookup on anything, even thought UDT does show DNS names of the Windows server.  So I'm not sure how this is working.  Assuming it does work like it should, in our case, a port will have a Cisco IP phone DNS name and a windows pc behind it then I'd expect both names to be listed.  Since this isn't teh case I would still expect the port details list under node details match device tracker port details for the same ports.  So I see 1 port with 2 IPs, the windows machine and IP phone, the windows machine dns is only listed and in the device port tracker veiw for that port I see the IPS only, no DNS names....not even for the windows one it does know about.

                If you need any information from me on this or need me to open a case I can do that for #7.  #4 I just need more clarification what the process is and if it's working like I'd expect it to do.  Thank you.

                BB

                  • Re: UDT - testing issues and requests
                    mavturner

                    michiganlaz, 

                    Unfortunately you are correct about the account levels. We are working on getting this fixed quickly. 

                     

                    Bryan,

                    #4 When you run Port Discovery, the first screen should include all SNMP enabled nodes. The only reason a node shouldn't be in that screen is because it doesn't have SNMP enabled.

                    #7 That is weird... best thing will probably be for you to open a case.

                    Mav

                    • Re: UDT - testing issues and requests
                      jpl3

                      >  We only monitor network to network ports 

                       

                      BB,

                      I'm curious how you did this - I can't figure how to do this short of editing each node after (subnet) discovery. I've got a few thousand nodes ,so that's not practical. Ideally I'd like a discovery option that could limit discover to interfaces running a link discovery protocol (CDP, LLDP).

                      If you have some method to buik remove the interfaces connecting end-stations (i.e. some SQL magic) I'd be very interested.

                       

                      thanks.. john