Comment 3 for bug 1828496

Hi,

I see you are using service(1) to reload haproxy. The service command has been written to manage SysV scripts, but nowadays it prefers calling systemctl when possible. You quoted that:

  After looking around for many days the issue was that "reload"
  operation created a new process without killing the old one.

and as haproxy ships with both SysV and systemd init files, this makes me think that service(1) may for some reason fail to call systemctl and ends up reloading haproxy using the SysV script, causing the double process. The first process would in this case keep running, serving the old certificate. However in the Xenial system where I tried to reproduce the issue `service haproxy reload` calls `systemctl reload` as expected. From /var/log/syslog:

May 23 09:39:42 xenial systemd[1]: Reloading HAProxy Load Balancer.
May 23 09:39:42 xenial haproxy[26047]: Configuration file is valid
May 23 09:39:42 xenial haproxy-systemd-wrapper[25945]: haproxy-systemd-wrapper: re-executing
May 23 09:39:42 xenial systemd[1]: Reloaded HAProxy Load Balancer.
May 23 09:39:42 xenial haproxy-systemd-wrapper[25945]: haproxy-systemd-wrapper: executing /usr/sbin/haproxy -f /etc/haproxy/haproxy.cfg -p /run/haproxy.pid -Ds -sf 26035

Do you happen to see two haproxy processes when a `service haproxy reload` fails to make haproxy serve the new certificate? Can you still reproduce the issue if you reload haproxy using systemctl instead of service? Can you please attach the tail of /var/log/syslog the next time you encounter the issue? Thank you.