Description of problem: After updating the proxy configuration, a service reload on stickshift-proxy causes the service to show as failed and systemd to terminate the haproxy daemon. The underlying issue is that systemd is confused by the daemon changing PIDs on reload. Version-Release number of selected component (if applicable): Fedora 17 and later. How reproducible: Always on affected Fedoras Steps to Reproduce: 1. systemctl status stickshift-proxy.service 2. systemctl reload stickshift-proxy.service 3. systemctl status stickshift-proxy.service Actual results: Service shows up as failed and is terminated Expected results: Service properly reloads and is not terminated. Additional info:
See Bug 721375 ; A similar issue was discovered with sshd and the resolution was to treat sshd as a "simple" service that does not fork into the background.
The PID file is no longer updated when haproxy is not run in daemon mode. Our procedures assuming it should change appropriately.
The following pull request creates a wrapper script for LSB compliance. https://github.com/openshift/crankcase/pull/474
Pull request accepted.