Applications all down at the same time

The most common application issues and how to resolve them are described.

You can determine the source of most fastcgi issues by examining messages in the log files produced by the GAS and/or the Web server. If the GAS is down, you can find some clues in:

  • GAS logs (see your appdata/log/<dispatcher-name>/date directory)
  • Web server logs
One reason could be a cron or a script that periodically shuts down and restarts the Web server. For example, in the Apache error_log file you can find messages like:
... [notice] SIGUSR1 received.  Doing graceful restart
... [error] [client xx] (4)Interrupted system call: FastCGI: comm with 
server "/opt/gas/bin/wrapper.fcgi" aborted: poll() failed
... [error] [client xx] FastCGI: incomplete headers (0 bytes) received 
from server "/opt/gas/bin/wrapper.fcgi" 

Apache received a SIGUSR1 signal. If you search through the error_log, you might have the message daily at the same hour. This is probably due to logrotate (see /etc/logrotate.d).

In order to archive the logs, logrotate sends a request to Apache to restart gracefully. This shuts down the fastcgi, the dispatcher and all Genero applications. You can check if this is really the cause of the issue by running this command (or an equivalent one):
/etc/cron.daily/logrotate --force /etc/logrotate.d/apache2

To solve, amend your logrotate to process when Apache stops, or at night, or save the logs of the previous day so that you can archive the logs without stopping Apache.