How do you all un-manage systems for patching/downtime/etc.?

Question

We currently essentially un-manage each node separately. I'm thinking of using Custom Properties and utilizing sites/environment to filter and utilize the unmanage task scheduler. However, I still think that's fairly clunky. Does anyone have a more streamlined approach to un-managed and/or re-managing large groups?

chadsikorra · Answer

The method I devised is a Powershell script that uses the Orion SDK/WSUS API calls. The script runs every hour, checks if a specific maintenance window is active, and toggles a custom property on the nodes called "inMaintenance" (True/False). When the maintenance window starts it toggles it to true. Then I have a check in the alert definitions to see if the "inMaintenance" property is set to true for the node. If it is set to true, then the alert will not trigger. The way the Powershell script knows when a maintenance window starts/ends is based on our WSUS group naming convention. Every WSUS group is named according to a standard "DD-HH.mm" format. So from a WSUS group name I can determine the exact day, hour and minute that a specific groups maintenance window starts (All of our maintenance windows are for a set period of time). I can then use the WSUS API calls from powershell to enumerate all members of the WSUS group and toggle their custom property using the SolarWinds API. Then it gets toggled back after the maintenance window ends from the same powershell script.

The whole system above works pretty nice because I don't have to modify the Powershell script that handles it all after it was configured. We just add servers to the correct WSUS group and stuff "Just Works". It also helped cut down on all the alert noise for the on-call person when patching was going on.

pcguywilson · Answer

We set up two different methods. Each with their pros and cons.

The first was using a script that is launched from an intranet site, there are multiple groups created you can select from and the script modifies the db directly. It changes the status of whatever nodes are in the chosen group to an unmanaged state. You can then remanage using the same method.

The second is what you're already considering which inst all that bad. Created a custom property called "PatchCycle" and created the various values. Then selected nodes that needed grouped and adjusted the property to display their cycle.

To put them into maintenance mode all at once, go to manage nodes and group them by the custom field patch cycle, then select the first check-box at the top, if there are additional pages of nodes it will ask you if you want to do all of them or just what you selected. Then you can click unmanage. You can the adjust the schedule or leave it be and re-manage when finished using the same method.

Also,

While you are on the manage nodes screen, you can add the column that displays your custom properties so you can see which cycle each node is apart of.

Hope that helps.

attachments.zip

janene16 · Answer

If you like, you can use SDK. Sample scripts shown here and here