Nginx failing at midnight

Hi,

I’m trying to track down a problem which is causing Nginx to be restarted at 00:00 every night.

I have a monit script set up to check the site periodically and restart nginx if there is a timeout greater than 8 seconds. The server is currently not being used live and I’ve never seen any response times over 6 seconds in New Relic. Usual response times are < 400ms.

The only clue I have to go on from the logs is that at the time it restarts, monit gets a 499 response from nginx which ive tracked down to being “client closed connection”.

[11/Sep/2009:00:00:13 +0100] GET / HTTP/1.1 “499” 0 “-” “monit/4.10.1” “-”

None of the other logs report anything around that time other than monit restarting nginx after the failure.

I’ve checked to see if there are any cron jobs running that could be affecting it but the only one that comes close is logrotate which sends a kill -USR1 command to nginx. Running this manually does not cause any problems with nginx. Daily cron tasks run at 6:25am as well so they can’t really be causing this.

Does anyone have any ideas why this would happen at midnight every night or where I can look to further diagnose the problem?

The server is configured to run Rails 2.1 on Nginx 0.7.61 and Passenger 2.2.4 with RubyEE 1.8

Thanks,

David.