monitor a cluster?

icky2000

I have a bunch of Exchange clusters. I've added the nodes individually and of course SW assigns the name of the node when they're added. That's fine. I thought maybe I could use the virtual IP of the cluster virtual server but that too resolves to the name of the local host even though that IP could move to the other cluster node. What's the best practice here? I'd consider leaving it as is with just the nodes but I also want to monitor Exchange functionality which is offered through the cluster virtual server IP but not necessarily the node IPs.

Find more posts tagged with

exchange

Disk

cluster

SNMP

sql

apm_2.0

discovery

virtual_server

localhost

monitoring

Accepted answers

All comments

freak523

We have a number of SQL clusters that we monitor. Basically we add the virtual IP in (Orion will pick up the machine name it is running on) Just rename the resource while keeping the cluster virtual IP and you will be able to monitor that resource regardless of which host it resides on. For us we just monitor the specific resources such as processes and disks that are associated with the cluster resource group.

-David

jrich

David,

I'd like to know how Solarwinds reacts when your cluster is failed over to the other node. Reason being is that if it were to fail over it would have to look at SNMP on the other node in order to monitor correctly. Have you been able to overcome this?

freak523

We haven't had any issues during fail-over of cluster resources. Since we are monitoring resources that are contained in the cluster resource group only they will move with the group. The host-specific information will update when solarwinds polls but it will not effect any other polling. The services and disks move over and report in SNMP without any issues, we actually use the host level information to help identify resource group failover and current node that it is residing on for alerting and reporting (useful in a 3 or 5 node cluster). Adding items such as memory and CPU can be useful since it will show the host information that the cluster resource group is living on at the time.

-David

robgarbee

I have a related question as to how to monitor a cluster environment. Much like the previous poster we have an Exchange cluster and we have want to monitor the two servers that support the cluster. We are able to monitor both servers with the default APM templates with the exception of the Information Store. What we are seeing is that while one server holds the Information Store function the other server will show that its Information Store is down. This is normal behavior however APM interprets this as a problem. I would like to see the ability to select the Information Store component and show that a down status is normal and show the application as green. Conversely when the Information Store component comes up the application would show red as this is not normal behavior. This would allow me to see that all the servers are running in a "normal" state even though one of the components are down. I hope I have explained my issue properly. Please let me know if you need more clarification.

-Rob