We have unique problem. We live in India where we have power outage for at least 10-20 mins - once or twice a day. We run a small demo server where I deploy "in progress" apps for my clients. We have been using and evangalising RoR since 8-10 months now in our organisation. I have mongrel configured to serve out the web apps and it dutifully stops and starts during system shutdown/reboot. The problem is when I have an unexpected power outage. (I know I should have a UPS/power backup and I will be getting one soon but the problem still cannot be ignored and a UPS is at best a temporary solution.) When power is cut off to the server, the 'mongrel.xxxport.pid' files remain in the respective log directories and prevent the mongrel servers from starting up. I have to manually delete them and restart the mongrels. Is there any way I can avoid manual intervention? (I could write a script to delete the pids and put it in run level 3 but that is pretty ugly.) This problem does not exist with any other server - apache/postfix/subversion..... and I do not think it should exist with mongrel either. Do let me know if I can help. I have done socket programming in 'C' before and have really smart people with ruby knowledge. Thanks, kc www.rknowsys.com
on 2007-04-02 06:59
on 2007-04-02 21:57
On Apr 1, 9:59 pm, kc <ruby-forum-incom...@andreas-s.net> wrote: > When power is cut off to the server, the 'mongrel.xxxport.pid' files > remain in the respective log directories and prevent the mongrel servers > from starting up. I have to manually delete them and restart the > mongrels. Is there any way I can avoid manual intervention? (I could > write a script to delete the pids and put it in run level 3 but that is > pretty ugly.) I've been running mongrel_cluster on Debian, and putting my .pid files in /tmp. Debian empties out /tmp on reboot, so this has worked fine. I should look in to what kind of housecleaning goes on with properly putting .pid files in /var/run, but I haven't.