Forum Discussion

Piotr_Lewandows's avatar
Piotr_Lewandows
Icon for Altostratus rankAltostratus
Feb 29, 2016

Monitor one VS pointing to multiple vHosts

Hi,

 

I wonder if my idea is correct or there is better way to configure LTM.

 

Scenario:

 

  • Multiple services using one IP:port and different FQDN (vHosts)
  • HTTPS used

Setup:

 

  • One VS with SSL offload
  • Local Traffic policy selecting pool based on Host header
  • As many separate Pools as vHosts - each with identically configured Pool Members (IP:port) but unique monitor assigned
  • As many monitors as vHosts, each using correct Host header in GET request

Result

 

  • Each vHost (separate Pool Member in separate Pool) has assigned monitor with correct Host header
  • Traffic is directed to correct Pool based on Host header content using Local Traffic Policy (or iRule)
  • Only service that is not working is marked down by monitor - not affecting other services under identical IP:port pair.

Is above correct or maybe there is some better way?

 

Piotr

 

1 Reply

  • Hi Piotr,

    personally I'm using a dynamic pool status page on every HTTP-enabled Virtual Server for health minitoring.

    The page will enumerate the currently selected pool and display the online/offline status of each pool member (this page can be used for PingDOM to monitor the health of each server). But the page has also two additional gimmics to access a specific node by using a.) a query string of

    ?f5action=select&node=[NodeID]
    (e.g for PingDom Monitoring) or by injecting a HTTP-cookie (via HTML buttons) to tie a specific browser session to a specific node (e.g. useful for developers to troubleshoot a single node).

    Compared to your outlined logic my status page approach wouldn't add additional overhead to monitor each pool and also monitor and access each single pool members. The status page information is dynamically generated based on just the currently selected pool.

    Note: Did send you an PM to show you my monitoring page in action... 😉

    Cheers, Kai