I am new to IM and I want to know that If a robot is installed on a server, do I need to add that server to the central net_connect probe to monitor it's availability?
First you need to decide on the criteria you want to use to determine whether something is available. This is one of those things that we leave up to the customer. Using net_connect is a very straight forward method of defining availability from a conceptual standpoint, but as johnhcw pointed out, if you are storing QoS data from cdm on a robot, you have the bases for also defining availability based on missing sample data.
That being said, the hub also tracks availability based on robot check-in. When a robot comes online it tells the hub that it should expect a check-in every few minutes. If the hub doesn't hear from that robot in a timely manner, then the hub raises an alarm to alert you to the fact that there is a robot that is not communicating. This can be another method to determine availability if you're looking for it from an alerting standpoint.
Yes ,for availability you need configure net_connect.
No, you do not need to do that, it's just one way to go about it. Considering licensing model, it actually doesn't make much sense to use it for that alone. You have your probes there that are constantly sending QoS data, for example disk or cpu usage. You can define your SLAs in such a fashion that it looks for missing samples in the data stream. I find this to be a much better indicator of availability than simply pinging a target or even doing a service test to a port. It is also easier to manage.
For some more information what David stated above, check out the following article.
"When a robot comes up and attaches to the hub, it sends out a robotup message which includes the next interval hub should expect another robotup message. This is based on the interval settings on the robot's controller GUI Setup tab >Misc "Hub update interval (in minutes)".
If the hub misses the heartbeat from a controller in approximately 1.5 times the heartbeat interval e.g., if the heartbeat interval is 10min, then in about 15min; then the hub does a robotcheckin to check if robot is active. If robot does not respond, then the hub generates a "robot inactive" alarm.
The severity of the alarm is controlled by a modifiable setting on the Robot tab of the Hub Configuration Tool."
using the heartbeat is also a valid option, I agree. Few things I'd point out, though:
Good points - thanks for the additions, John.
Forgot to mention that as for using QoS interval to measure availability, you also need to take into consideration how your robots are configured to handle interruptions in network connectivity. It can either spool qos messages then or drop them. If you're responsible for the network part also, you'll likely want them drop if considering purely from SLA side of things.
Thank you all for the answers, helps a lot!
Retrieving data ...