Bug #15182
openChanging backend port - status remains down
0%
Description
I noticed that if you enable basic health checks on a backend server, in my example port 443 the health check succeeds.
Later on if the port changes and you update the configuration to reflect that, the healthcheck fails. In my case reviewing the haproxy_stats page shows that its still checking port 443 even though the configuration is set for port 80.
I also receive a new 'warning' from the package when i attempt to apply my changes
I did notice that if the backend server health check is DOWN already before you make the health check modification this is what causes the problem.
1. first have a successful health check.
2. Change backend application to a different service port. In my case from port 443 to port 80
3. Update the health check. If you check the status it should be down
4. Attempt to apply config. You will see the warning but that prevents the changes from going forward.
I tried with 23.05 and my problem doesn't exist on that version.
Files
Updated by Mike Moore 10 months ago
The workaround is to delete the backend server and recreate
Updated by Danilo Zrenjanin 9 months ago
- Status changed from New to Confirmed
I can confirm this behaviour.
[NOTICE] (69051) : haproxy version is 2.9-dev2-80cef0c [NOTICE] (69051) : path to executable is /usr/local/sbin/haproxy [WARNING] (69051) : config : Server bacnendList_ipvANY/backendSense is DOWN, changed from server-state after a reload. 0 active and 0 backup servers left. 0 sessions active, 0 requeued, 0 remaining in queue.
Tested against:
haproxy-devel 0.63_2