Hi everybody,
First of all, we have been using haproxy on our cluster for about four
years now, We have never had any downtime caused by haproxy and performance
has always been excellent. So a big thank you for this excellent piece of
software!
For a while now (unfortunately I cannot pinpoint the exact version) I am
seeing strange service down notifications in my logs. It is a backend used
to monitor MySQL replication health, the backend itself is very simple:
backend db01_replication
mode tcp
option httpchk
server db01 server_ip:3306 check port 9201 inter 5000 fastinter
2000 downinter 2000 rise 2 fall 3
Port 9201 on the server_ip contains a simple xinetd script which connects
to the mysql server and checks the replication status. In case everything
is ok the xinetd script returns status 200 ( echo -e "HTTP/1.1 200
OK\r\n"), otherwise it returns 503.
This works ok 90% of the time, but occasionaly I see strange 'down'
notifications like this:
Dec 28 18:20:32 localhost haproxy[29371]: Server db02_replication/db02 is
DOWN, reason: Layer7 check passed, code: 200, info: "OK", check duration:
44ms. 0 active and 0 backup servers left. 0 sessions active, 0 requeued, 0
remaining in queue.
Service is DOWN because check passed?
Can anyone helpt me to troubleshoot this issue?
Regards,
Reinout Verkerk
Directeur
*Trilex Online Assessment*
Burgemeester Roelenweg 33
8031 ES Zwolle
T: 085 - 401 22 00
M: 06 - 553 76 034
www.trilex.nl
