If I setup a PRTG Cluster with 1 or more standby nodes, how will the users interact with the environment if the primary or master node fails?

Scenario:- My users and remote probes connect to my PRTG server via NMS.XXXXX.COM, which in normal operation will present them to the Master node. If the master node fails, or is taken off-line, (a) How will the user be able to connect, and (b) how will Remote Probes be able to present their data for processing?

Thanks Ian

Article Comments

Dear Ian,

Remote Probes can only connect to the master node in a PRTG Cluster, and only the Master Node is fully aware of these Remote Probes (so even if you would manually "switch" a Remote Probe to another Failover Node PRTG Server, while the Master Node is down for a short time, the Failover-Node would not show the Remote Probe). Remote Probes do continue to monitor and also cache their collected data if they lose connection to their PRTG Server, so once the connection is re-established they will send all the data to the PRTG Server. For the users loging into your PRTG Cluster, though we would recommend to redirect them to the webinterface of the Failover Node while the Master is down.

best regards.


Nov, 2011 - Permalink

I must be misunderstanding the concept. I am looking for a way of maintaining our capability to deliver a "service" in the event of a server failure, so assumed that Clustering would provide that capability. From what has been described above, Clustering will only provide a method by which locally monitored devices (i.e. attached to the "Local Probe" of the server) are accessible, whereas in my case 90% of the devices are all monitored by remote probes.

Is it not possible to run PRTG in some form of Active/Active clustered enmvironment?

I'd be happy to have a phone call with the team if it would help to understand my requirement. Ian


Nov, 2011 - Permalink

We do understand your request/scenario, but Remote Probes cannot be clustered in the moment. Only the sensors under the "Cluster Probe" are clustered and so monitored by each node of a PRTG Cluster (essentially then by each Local Probe).


Nov, 2011 - Permalink

It seems that clustering is only working fully when you are not having a large installation.

Please implement full clustering feature that can scale up with probes.


Dec, 2011 - Permalink

I am also in this same situation, I have over 9 remote probes providing for over 18,000 monitors. So what your saying is if I lose my master node my ability to alert and provide active monitoring is lost and you have no way to provide me a solution to resolve it.

I guess this means I need to start looking for a new product.


Aug, 2014 - Permalink

We understand well the unpleasant situation for those of you who use Remote Probes heavily and still need a fail-safe cluster setup.

Currently, there is only one solution for a fail-safe monitoring in different locations: At each location, install two PRTG core servers and configure them to run as a two-node fail-safe cluster. Using the Enterprise Console, you can still connect to all of your individual installations and see all data in one console.

However, we know that the need for remote probes that are connected to several cluster nodes is increasing. So, we're definitely looking into a solution for this. But as always, I cannot give any dates yet. We're still in research for this.


Aug, 2014 - Permalink