Forum Discussion

Eric_Singer's avatar
Eric_Singer
9 years ago

Collector configuration template

Hi,

One of the challenges I've had with collectors is getting them sized right based on the load we have.  It seems like whenever we run into performance issues, we're asked to deploy more collectors, yet the underlying collector doesn't show signs of high CPU or high memory.  I'd really like to see a template for basically scaling a collector up, or even sizing a collector based on the average devices you want to run per collector.  IN other words, in the console have a drop down where the template size is chosen and applied.

For example:

Small = <=100

Medium <=250

Large <=500

X-Large <=1000

For each config size, there would be a recommended minimum CPU / Memory config for the collector, and when the template would be selected, it would automatically tweak things like outstanding task queue, timeouts, parallel tasks, etc.  

 

  • Eric_Feldhusen's avatar
    Eric_Feldhusen
    7 years ago

    We really like the small/medium/large presents for the Collector templates, but we've noticed in our usage, that for us, there's two sets of additional Collector configuration changes we make based on whether the collector is doing primarily a large amount of purely snmp queries or if it's doing a large amount of pings and tcp pings to specific ports, so we have to "tailor" our collectors based on the devices in use on that collector.

    If it was possible for us to add to the LogicMonitor supplied Collector templates of small/medium/large with additional "templates" specific to our instance of LogicMonitor, it would make the collector deployments extremely fast.

    At the moment, we have to spend time after each collector installation doing large copy & paste of collector configuration templates.

    There are other ways to do this with configuration engines like Ansible, Chef, Puppet, but it makes sense to keep that type of information within the LM interface.