Load Balancers not unbinding - Bug?



  • Hi,

    We've found that a load-balancer virtual server is configured to point to a pool  and then removed again the binding/virtual server isn't actually released.

    Steps:

    • Configure Pool (example: 192.168.0.63,192.168.0.64)

    • Configure Virtual Server (example: 10.1.1.1:80)

    • apply the configuration

    • remove the virtual server

    • apply the configuration again

    At this point, if a connection is made to 10.1.1.1:80, it will still forward the connection to the pool, despite having been removed.

    If one power-cycles the box then the configuration is correct again.

    I've checked http://redmine.pfsense.org/ and couldn't see any sign of this issue anywhere, is this the correct place to report bugs?

    Secondly, is there a way of restarting the load balancer daemon manually in order for it to correctly unbind/rebind ip and ports again without restarting the entire box?

    Thanks!
    -Rob


  • Rebel Alliance Developer Netgate

    Some things in relayd require a full reload (kill/restart), but some require only re-parsing the config. Most of the changes made in the GUI only trigger that faster refresh, to minimize downtime.

    You can always restart the service from Status > Services which will fully reload (stop/start) the daemon.


Log in to reply