To assist with batch node operations, NMIS includes a little script for importing nodes from a CSV file. From version 9.1.1G onwards there are also more fine-grained tools available, which are described on the page titled Node Administration Tools.
IMPORTANT: The import_nodes.pl script was updated on 10 Dec 2020 to better handle node activation which was causing problems with nodes. If you are using this tool you should update from the NMIS9 GitHub repository, the links are below.
The minimum properties you need to have to add a device to NMIS are, name, host, group, role, community, roleType and netType. Technically you can use defaults for group and role and the name and host maybe the same, so the absolute minimum is host and community. This sample CSV includes the full five properties and if required you can use additional ones.
Use the activated.XXX field to activate for NMIS, opConfig, opEvents, you can activate for NMIS but not opConfig for example.
To load these devices into NMIS9 run the following command:
This will take the CSV file and will look for the existence of the node based on the name/node_uuid. If the node exists, it will override its properties for the properties specified in the csv.
By default, NMIS will be run in simulation mode. At the end, you will able to see if the node will be created or update. As an example output:
If you are ready to run the command and apply this changes into NMIS, you should use simulate=f. As an example output:
Once you have added nodes or modified nodes an NMIS Update is required which you can run for all nodes or run for a single node or simply leave it until the next scheduled update for that node is due (default every 24 hrs).
If you are running an update for all nodes it may take a while to complete. The following command shows how force an update for one node at a time (a good way to distribute the load). You can also schedule an update for all nodes by removing the job.node argument.
To run an NMIS update for a single node, optionally with debug which will result in debug files in /tmp/
If adding a large number of devices it might take sometime to complete the adding. This is because the first time a node is added to NMIS it needs to create all the RRD files for performance data, this only takes a few seconds per file, but each node may have 10 RRD files or more, so this compounds to a large number of seconds very fast when adding 1000's of devices at a time.
You can check how long the script is going to last for each node using the following argument:
The output show be something like:
Import nodes from the primary into the pollers (opHA)
We can use the primary to import nodes to the pollers, indicating the cluster_id in the header:
The cluster_id can be checked:
- Using opha-cli tool:
In the opHA GUI:
The nodes will be transferred when the opha cron job runs, it can be called manually:
More information in the opHA documentation opHA 3 Redundant Node Polling and Centralised Node Management#Managingremotenodes