Control: retitle -1 rt4-fcgi: init script doesn't wait for database server Control: tags -1 + moreinfo
On Mon, Aug 03, 2015 at 12:44:40PM +0200, Marco Pessotto wrote: > Package: rt4-fcgi > Version: 4.2.8-3 > Severity: important > > Dear Maintainer, > > installing request-tracker4 with rt4-fcgi in a systemd environment (as default > install) appears to lead to a problem at boot time. > > It looks like that the service is started when the database is not ready: > > ● rt4-fcgi.service - LSB: request-tracker4 FastCGI daemon > Loaded: loaded (/etc/init.d/rt4-fcgi) > Active: active (exited) since Mon 2015-08-03 08:50:05 CEST; 3h 33min ago > Process: 434 ExecStart=/etc/init.d/rt4-fcgi start (code=exited, > status=0/SUCCESS) > > rt4-fcgi[434]: spawn-fcgi: child spawned successfully: PID: 470 > rt4-fcgi[434]: spawn-fcgi: child spawned successfully: PID: 471 > RT[470]: [470] DBI connect('dbname=rtdb;host=localhost','rtuser',...) > failed: Can't connect to loca...ine 105. > RT[471]: [471] DBI connect('dbname=rtdb;host=localhost','rtuser',...) > failed: Can't connect to loca...ine 105. > RT[471]: [471] Since your configuration exists > (/etc/request-tracker4/RT_SiteConfig.pm) but is not ...nything. > RT[470]: [470] Since your configuration exists > (/etc/request-tracker4/RT_SiteConfig.pm) but is not ...nything. > > Without human intervention, a rebooted machine responds with 502 to any > request > to the RT urls. > > Restarting the service after boot fixes the situation, but it's obviously not > optimal. Hmm, I don't believe that this is a systemd-specific issue, although it's obviously exposed by the way systemd works. I believe that adding the following lines to /etc/init.d/rt4-fcgi just below "Default-Stop" should fix the issue assuming that RT is configured to use a database server on the same host, of course): # Should-Start: mysql postgresql # Should-Stop: mysql postgresql Please could you let me know if this is the case? Cheers, Dominic.