11 Replies Latest reply on Nov 9, 2012 2:54 PM by rharland2012

    Question regarding Polled Status of UPS

    jklandes


      Ok, this might seem silly, and I am a bit new to NPM, but we have a few UPSs that are sending traps because they need their batteries replaced....

       

      Question: Why doesn't the [polled SNMP] node status for these UPSs show as being in warning or critical state on my system? One in particular is an APC smartUPS 450. It's sending traps about the battery, but the node status still shows green and healthy. I would think the battery problem should be in the MIB and casue the stauts of the node to at least go to warning? Am I completely missing something? I configured the traps to send us alerts, but there is no option to change the status of the device, only the interface???

        • Re: Question regarding Polled Status of UPS
          rharland2012

          You can leverage Universal Device Pollers to get a richer set of information pulled from the UPS. Click on the 'Content Exchange' tab at the top of the Thwack page, choose Universal Device Pollers, and search for 'APC'. I think you'll find something you can populate your node details page with that will also allow you to create some alerts.

            • Re: Question regarding Polled Status of UPS
              jklandes

              I think I may have closed the question…. No way to respond in the thwack web interface so I’ll try replying to this email…

               

              Thanks so much. I found & installed the APC universal poller, assigned to the devices, updated the node detail view and am seeing the UDP status information. For a representative APC that needs batteries I can see the upsAdvBatteryReplaceIndicator has a value of batteryNeedsReplacing, so we’re getting a valid reading back, but the node status is still green. Going into Advanced Alerts I don’t see how I can set up a trigger that looks at this specific value, so I’m left with how to change the node status to warning or critical based on a UDP variable value. Do I need to customize the UDP to change the node status based on the UDP specific attribute value?

               

              Thanx!

                • Re: Question regarding Polled Status of UPS
                  rharland2012

                  You just have to log into thwack to be able to reply, etc.

                  That's a great question - I'm sorry I glossed over what you were truly after! I haven't yet attempted to change node status with a UDP alert. People *have* asked for this feature in the past, I believe. This would be a great feature request! I'll dig a little bit and see what I find. Good luck to you!

                    • Re: Question regarding Polled Status of UPS
                      jklandes

                      I think I have it… I’ll post results in a few….

                       

                      Thanx!

                        • Re: Question regarding Polled Status of UPS
                          rharland2012

                          Did you ever have any luck on this one?

                            • Re: Question regarding Polled Status of UPS
                              jklandes

                              Thanks for following up on me ☺

                               

                              Well, while you CAN create an alert trigger based on the value from a UDP it does not appear that you can include basic node information like the caption, etc., in the email (or in the trigger, for that matter). The only data you can reference anywhere in the alert rule is what’s included in in the particular UDP, so it’s less than useful for my problem.

                               

                              The other thing I verified is there is no support in NPM to change the state of a device via an alert action (this was in a forum thread about this from 2010 that ended with “moving this to feature request forum”. So, even though you can clearly see that the battery needs replacing in the node details, you can’t turn the node red so it stands out in the console. You can’t change the status of a device from a trap action either.

                               

                              I have a trap alert configured for several traps from the PowerNet-MIB (what APC UPSs use) and it’s notifying us about the dead batteries. It sends an email which eventually repeats every 6 hours, but that’s the only indication that there’s any kind of problem – otherwise the UPSs with bad batteries all stay green in the console (grrr).

                               

                              Thanx!

                                • Re: Question regarding Polled Status of UPS
                                  rharland2012

                                  Thanks for the info - I'm going to resubmit as a feature request. This would be super-valuable! Good catch..

                                    • Re: Question regarding Polled Status of UPS
                                      jklandes

                                      I don’t think adding a function to change the status of the node from an alert action is right approach and it’s probably why the feature isn’t provided today. The real issue as I see it isn’t so much setting the device status when the alert is true (you can configure a trigger to do this fairly easy), it’s how to correctly set the new status later when the condition clears. The status of the node is really an aggregate of multiple inputs, not just the UDP status, so you could not simply add a “Set Node State Green” action to the Reset Action of the alert definition – a completely different trigger (more than one, probably) would be needed to set it correctly based on all the other conditions that would go into determining this node’s actual status.

                                       

                                      What makes more sense to me is to enhance the UDP to support optional alarm states for the UDP based on values for the data returned within the UDP - the UDP would essentially have its own status as a part of its definition. Now think of the UDP like it’s a dependent of the node – just like a port. Port statuses are merged/integrated onto the status for their parent router or switch. Why not treat the UDP the same way? In fact the same red box you see for a port alarm in the console would work nicely for the presentation element – just enough that it stands out. I think this would be less clunky than the Alert-based approach since it would be integrated within the UDP definition and possibly be faster to release from a software development perspective since it would leverage existing code to handle port alerting in the web interface .

                                       

                                      Thanx!

                                    • Re: Question regarding Polled Status of UPS
                                      byrona

                                      You should be able to send basic node information in Custom Poller alerts, we have it configured to send the Node Name as well as custom properly information and it's working.