I have a HAproxy instance with one backend and two mysql servers. My question is Can I set for each server in backend different healthchecks? It could be perfect to set mysql-check to backend server and tcp-check to main server. I didn’t find this in official docs.

Here is backend’s config

backend db-backend
        #option mysql-check user ###
        option tcp-check
        tcp-check connect port 10000
        tcp-check send GET /check-status HTTP/1.1rn
        tcp-check send rn
        tcp-check expect rstring "200"
        
        server data1 10.0.0.20:3306 maxconn 500 fall 2 inter 2000 rise 10 backup
        server data2 10.0.0.20:3307 maxconn 500 check fall 2 inter 2000 rise 10 

I had a mysql-check, it’s commented in config, it worked well, but now I have task to check not only health of mysql instance. One instance is backup and they have master-master replication. So, when main server rises, I need to be sure that replication not failed. I made tcp-check for this, it checks status of slave_io_running (Yes), slave_sql_running(Yes) and seconds_behind_master(0) and respond with code 200 OK. And only then main database set to status UP.
If main server failed, parameter slave_io_running on backup server will change to “connecting”, so my check will fail and I loose whole backend.