Forum Discussion

Hoang_Hung's avatar
Oct 24, 2022

Heath check range service about 1000 port

Hi all 

I have a service need monitoring service ( pool LTM) with range 4500-4600. How do you know solution for it ?

Thanks

Hoang Hung

  • Hi Hoang_Hung, you can approach this a couple different ways.

    The first is to create a monitor for each port you need to make sure is available, and then attach all the monitors to the pool as alias port and make sure they are all successful. Something like this (unrelated monitor and pool details removed for brevity):

     

    ltm monitor http http_8080 {
        defaults-from http
        destination *.8080
    }
    ltm monitor http http_8081 {
        defaults-from http
        destination *.8081
    }
    ltm pool my_pool {
        monitor http_8080 and http_8081
    }

     

    Another option is to create an external monitor using a command line tool like curl/netcat, or writing your own perl/python script (existing versions on BIG-IP to be clear) to probe all those services and make sure they are up and then report back to the monitoring service appropriately. For example, you could use netcat to probe the services in a loop and exit on any "Offline" response. Challenge here is timeliness with netcat, with one hundred services this makes your interval a concern unless you're going to thread your scripts, which adds a lot of complexity on an EAV and probably not advised. that said, the netcat option I was toying with:

     

    me@mac ~ % nc -v -z -w 0 10.0.4.20 8079 &> /dev/null && echo "Online" || echo "Offline"
    Offline
    me@mac ~ % nc -v -z -w 0 10.0.4.20 8080 &> /dev/null && echo "Online" || echo "Offline"
    Online
    me@mac ~ % nc -v -z -w 0 10.0.4.20 8081 &> /dev/null && echo "Online" || echo "Offline"
    Online

     

  • The way we did this to monitor 4000 ports was monitor 1 or 2 with a monitor and that's it.
    We found that when creating that many vs's and pools thrashed the BIG-IP.
    But actually the breakig point was the monitoring tool that couldn't pool that many vs's / pool's quikc enough!
    But i would recommend talking to you software engineers, we had it documented that if one port goes down they all would, so monitoring 1 or 2 ports was just as strong as monitoring all of them.