HTTP monitor on the FQDN name of the node not the IP

[jda]

HI,

We have a Pool of HTTPs nodes that only reply if you come with the correct FQDN (let’s say server01.domain.com and server02.domain.com for the example)

Problem is that the included HTTP monitor makes the queries on the corresponding IP (10.0.0.1 and 10.0.0.2) and the server reject the HTTPs query because he only accept request on the a single host header, the FQDN.

I know I can force the “host_header” on the HTTP monitor, but unfortunately, there is no way to have a common host header on the backends.

Is the only solution a custom monitor? Did anyone already faces such a problem and have a solution?

[ldarby · Accepted solution]

 Hi jda,

You would need to use a custom health monitor for this.  Here is a script one of my colleagues already wrote:

#!/bin/bash

export $(echo "[email protected]" | sed -rn 's/--([a-z_]+=[^ ]+)/\1/gp;')
exec curl -sS -D - http://$node:$port$path | head -1 | grep -q '^HTTP/1\.. [234]..'

This uses curl with the pool's node name, which should send the appropriate Host header.

Regards,

Laurence

[jda · Accepted solution]

Thanks for the help.

We went the python route and I modified the "HTTPS_port-monitor.py" from Juergen Luksch to manage the case and get a little more control.

Working great for the moment