High-Latency link between minion and master

Hi,

We have a site that is on the other side of the planet in terms of geolocation in regards to the “master” OpenNMS server. We have been experimenting with using minions and was thinking of how we could utilize the minion in geographic places which are located far away from the master server.

Because we currently have an issue with monitoring devices which are located far away, which is generating a lot of “nodeDown” events which is simply because the response times are sometimes absurdly long.

Could we for example, place a minion in various geographic locations and somehow set it so that the link between the minion and master server is very generous in terms of timeout etc. So the minion would poll the devices in the far off geographic area and then report back to the master server but since the minion is now suffering from the same latency, the link between the master and minion would have to be tweaked so that the latency is not an issue.

Could this be done?

Yes. This is by design and I would probably use the AMQ vs Kafka broker. You can tweak the JMS communication to handle the latency. I would also probably setup an external AMQ broker vs the embedded broker.

Might also consider setting up a broker local to the Minions and have a broker to broker link between the remote sites and the NMS.

Thanks!

i don’t quite follow. How do you mean it’s by design? Is the minion created for occasions when there is high-latency monitoring involved?

Do you have any example of the tweaked JMS communication? And can this be set per-poller or location specific? Because we want our other minions to act normally.

Really interesting stuff this, would love to hear more of your thougths :slight_smile: