TCP health check port range failover

Help!
1

I am curious on what failover looks like with this configruation.

backend aus-back
balance roundrobin
option redispatch
server ct1 11.11.111.111 check port 6020-6023
server ct2 22.22.222.222 check port 6020-6023

Will this set up upon detecting a downed port, only failover that port to the other server in the cluster? Or will it failover the entire server to the alternative on the detection of a downed port?

As a follow up question…Is there any information on the resource consumption of a wide range of tcp health checks? say something like “check port 100-30000”. Would this be a huge resource hog on HAproxy or is this feasable pending the hardware capability. I plan to do testing on it soon but I dont have the equipment on hand yet.

2

It’s been a while since I confirmed so if the behavior has changed my apologies… But I believe it will check only one port in your example… Probably the first in your example or maybe it’s defaulting to the bind port…

I’d expect something more like this for TCP health checks:

 tcp-check connect port 6020
 tcp-check connect port 6023

Then the behavior I’d expect is it to fail the whole server should either port be down.

Again apologies if I’m missing something…

4

Ok i figured it would have been too good to be true. Ill just look into scripting options then. Thank you for your input.