Server line disappears from HAProxy config

[blueh2o]

I'm having a problem where occasionally I go to access a particular backend, and I get a 503 error. I go to look at the HAProxy config, and the server statement for that backend is missing:

backend Zoneminder_ipvANY
	mode			http
	id			106
	log			global
	timeout connect		30000
	timeout server		30000
	retries			3
	option			httpchk OPTIONS / 

If I restart the service, the line reappears:

backend Zoneminder_ipvANY
	mode			http
	id			106
	log			global
	timeout connect		30000
	timeout server		30000
	retries			3
	option			httpchk OPTIONS / 
	server			Zoneminder 192.168.1.13:443 id 107 ssl check inter 60000  verify none 

I am baffled as to why this is happening.

[PiBa]

@blueh2o
Is the server defined with a dns name?

[blueh2o]

Yes, it is.

[PiBa]

@blueh2o
Then i suspect the php code (sometimes) fails to resolve that name to the proper IP when the config was generated. Can you check when the configuration was generated and if 'something' special was going on at that time? Perhaps the dns server used wasn't reachable? Or it was during a reboot while not all services were running yet?

When the name cannot be resolved the server is skipped as otherwise haproxy would fail to start. Can you perhaps configure the dns-resolver settings of haproxy itself on the settings tab? That should get the server-name instead of the IP in the configuration and allow haproxy to resolve the name by itself at runtime.

Might also need to add this in the advanced global textbox setting:

defaults
    # never fail on address resolution
    default-server init-addr last,libc,none

That way it should avoid the situation above. It would still return the 503 while dns doesn't work, but when dns starts working haproxy would resolved it by itself.

[blueh2o]

Thank you for these suggestions. I'll try them.