© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
e3ace39
(Get the code!)
Wily is using the systemV init scripts and is not affected by this bug.
Trusty is using upstart which lacks a reload function currently.
Changing the logrotate postrotate script on trusty to call 'service freeradius reload' will not work without explicitly creating a reload function in the upstart job. The patch as included in comment #2 will kill the parent of freeradius (which is the upstart shell script) and will attempt to restart it but the original freeradius process is still running and no longer managed by upstart. When this occurs, upstart will show that the freeradius service is not running and will not be able to start it because the port is in use by the un-managed freeradius process.
Next action:
- Investigate adding a reload handler for the upstart job. This will need the PID for the freeradius process that was started so that it can send a HUP signal as seen in the systemV init job. In the case that the PID file is missing then the reload should do nothing and exit cleanly as the lack of PID file would indicate that freeradius is not running; we would not want to restart the service as this would have the affect of starting a stopped service. Then change logrotate script in trusty to call 'service freeradius reload' and consider removing the redirect to /dev/null.
I have run out of time today but have left these notes so that we can pursue it further.