For years my group has been a Solarwinds and Intermapper NOC. Around the beginning of the year we decided to move towards Solarwinds exclusively as we liked the direction and product offerings. Fast forward to today and we have a running test environment with NPM 9.0 SP1 so that we can form a strategy for moving off Intermapper. As things stand now we use Solarwinds purely for performance data as Intermapper handles all our alerting.
The point of this long winded question is to ask the hard-core Solarwinds users how you deal with the following:
1. It would seem that the one thing missing from the universal device poller is the ability to create templates or groups of probes. If I have 15 probes built that make up the things I would like to monitor for all my Cisco 3750's it would be great to just add the node and then assign that template/group to the node. Anyone figured out how to do this through SQL or via another method?
2. One of the real things we are struggling with is alerting. Intermapper makes it pretty easy to setup users/groups and assign them to probes. So, what strategy have you come up with to deal with the following:
a. Disabling of alerts for downstream devices? i.e. if a router tanks I do not want 50 messages that all the switches connected are out also. On the surface this sounds simple but when you consider the number of nodes we monitor and the number of custom pollers/alerts per node it can get a little crazy. I think we might be over thinking this one and there is a simple solution out there.
b. Do you have a master condition for alerting? From time to time we have power outages or other crazy things (people trying to steal cable from a live power substation) that happen where we want to turn off all alerting. It would seem to make sense that the first condition of an alert would be to check for a master value to make sure alerts should be going out. Assumption here is a field in the db which can be easily toggled. Curious if anyone has this implemented this and how.
c. Looking at everything we have about 25 people from 5 different groups that get alerted. Those people are in certain groups based off the nodes they are responsible for. For each node, depending on what is monitored, there is between 1 and 10 priorities that indicate when someone should be notified. The priorities are not consistent between groups and often look like "between 8am - 5pm, Monday - Friday". With there not being an address book or the ability to create notification groups how do you handle you diverse notification needs? It looks like we can solve part of this through a custom field or two but we are thinking that who belongs to what group might be handled via a list serv or something along those lines. This would keep us from having to edit x number of alerts when an employee comes or goes.
I could probably go on and on but I do not want to press my luck much further.
Appreciate any feedback you might have.
Maybe I am misunderstanding what you are trying to do for #1. Each poller is a "template". You create the poller then assign it to 1 or more nodes. And pollers can be grouped. Is there something specific tyou are trying to do that you can't?
I was swinging for the fences on this one.
I would like to say here are the 15 custom pollers I have for a specific device, say a cisco 3750. With all the pollers grouped, and called Cisco 3750 Pollers, be able to assign that group to one or more nodes.
When you have 75 Cisco 3750's and 15 probes for that class of device it can be rough going through and assigning each probe. Luckily solarwinds does allow us to select a group of nodes via grouping.
Regarding alerting, there can certainly be some challenges. As far as I've found, there currently isn't a good way to disable alerting for downstream devices. I remember reading a SolarWinds white paper or something in an admin guide at some point regarding this, but you basically had to configure a different alert for each device. We have almost 3000 nodes, so this wasn't an option for us. I have seen other threads in the forums requesting this functionality. I think it's on SW's radar, but I can only imagine how much fun it will be to code that kind of a feature.
I haven't implemented a master condtion but I would think you could set a master condition to be checked before triggering an alert, just add the condition to your trigger conditions and set the condition to trigger the alert when all of the conditions apply.
For setting alerting by group, it is a bit of a pain to set up initially. You could use custom properties to add devices to a specific alert and then set your time of day conditions for each priority. If you have distributions groups for the groups receiveing the alerts, you could set your trigger action to send an email to that distribution group's email address. That way, you won't have to change the alert as group membership changes.
I think you hit it right on the head with the word "initially". As we move to NPM exclusively there is going to be a large amount of upfront work which we can not avoid.
I am also pretty much sold on an external distribution list. Seems like the easiest way to deal with the complexity.