The solution is to directly delete them from Trac’s SQLite backend. There are several type of comments: Plain ones and comments that describe a change to the ticket’s properties. I’ll deal with only the plain comments, but the methods I describe should also be applicable to the other kinds.

First things you should note down the ticket number and the comment number. The comment number appears in the URL of the permalink to it. It is the number that is pointed by an arrow in the screenshot above, it will be displayed when you hover over the circled link. Now in the shell, cd to the db directory of you trac installation and do:

sqlite3 trac.db "delete from ticket_change where ticket=<TICKET NUM> and field='comment' and oldvalue=<COMMENT NUM>"

For example for the ticket in the screenshot:

sqlite3 trac.db "delete from ticket_change where ticket=12 and field='comment' and oldvalue=3"

You can delete different kinds of tickets by using other values in the field column, such as keywords and summary. Another useful column is the author column.

After searching a bit I’ve found out that cron can run tasks on startup using the special @reboot keyword instead of the normal time fields. So edit your crontab and add the following line:

@reboot /usr/bin/tracd --daemonize --pidfile=~/run/tracd.pid --port=PORT --hostname=HOSTNAME -s TRAC_ENV

Just replace PORT, HOSTNAME and TRAC_ENV with the appropriate values for your environment, and make sure you got a run/ sub-directory in your home folder (or change the pidfile value).

To stop the server just do:

kill `cat ~/run/tracd.pid`

While there is no straight way to restart the server (like /etc/init.d/tracd restart), it’s a good compromise for dropping root privileges.

for i in `seq FROM TO`; do trac-admin TRAC_ENV ticket remove $i ; done

Replace FROM and TO with the first and last ticket numbers of the range accordingly. TRAC_ENV should be the path to your project’s environment.

For example:

for i in `seq 9 32`; do trac-admin www.openyahtzee.org/ ticket remove $i ; done

The next problem I faced was how to start the Trac automatically upon startup. The solution was to use an init.d script for stating Trac. After some searching, I didn’t find an init.d script for tracd that were satisfactory (mostly poorly written). So I went on an wrote my own init.d script for tracd.

#!/bin/sh
### BEGIN INIT INFO
# Provides:          tracd
# Required-Start:    networking
# Required-Stop:     networking
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: Start the tracd standalone Trac web server.
### END INIT INFO
# (C) 2008 Guy Rutenberg <http://www.guyrutenberg.com>
 
## Options you should probably change ##
HOSTNAME=127.0.0.1 # to which hostname should we listen
# If you only want to serve one project keep this variable
# empty and set the PROJECT_ENV variable 
ENV_PARENT_DIR=/home/guyru/trac.guyrutenberg.com
PROJECT_ENV=
PORT=9090
# add any additional options (such as authentication) here. If you only have one
# project you should probably add -s here
ADDITIONAL_OPTS=
 
## Options you should probably not change ##
DAEMON=/usr/bin/tracd
NAME=tracd
DESC="web server"
PIDFILE=/var/run/$NAME.pid
SCRIPTNAME=/etc/init.d/$NAME
SSD="/sbin/start-stop-daemon"
 
test -x $DAEMON || exit 1
 
set -e
 
. /lib/lsb/init-functions
 
DAEMON_OPTS="--daemonize --pidfile=$PIDFILE --port=$PORT --hostname=$HOSTNAME $ADDITIONAL_OPTS"
if [ -n "$ENV_PARENT_DIR" ]; then
	DAEMON_OPTS="$DAEMON_OPTS --env-parent-dir=$ENV_PARENT_DIR"
else
	DAEMON_OPTS="$DAEMON_OPTS $PROJECT_ENV"
fi
 
case "$1" in
  start)
	log_daemon_msg "Starting $DESC" $NAME
	if ! $SSD --start --quiet\
	--pidfile $PIDFILE --exec $DAEMON -- $DAEMON_OPTS ; then
            log_end_msg 1
	else
            log_end_msg 0
	fi
    ;;
  stop)
	log_daemon_msg "Stopping $DESC" $NAME
	if $SSD --stop --retry 30\
	--pidfile $PIDFILE ; then
	    rm -f $PIDFILE
	    log_end_msg 0
	else
	    log_end_msg 1
	fi
	;;
  restart|force-reload)
	$0 stop
	[ -r  $PIDFILE ] && while pidof -x $NAME |\
		 grep -q `cat $PIDFILE 2>/dev/null` 2>/dev/null ; do sleep 1; done
	$0 start
	;;
  *)
	echo "Usage: $SCRIPTNAME {start|stop|restart|force-reload}" >&2
	exit 1
	;;
esac
 
exit 0

The script begins with a configuration section. You should go over each variable and make sure its value suits you need. The first part is configuration variables you are more likely to change. It includes:

• HOSTNAME – This tells tracd on what hostname to listen. From my experience setting it to anything other than 127.0.0.1 or localhost will cause it to operate on all available hostnames. If you use tracd in conjunction with mod_proxy (this is usually the case if you have Lighttpd) you would want to leave the hostname on 127.0.0.1 so the tracd server won't be accessed directly.
• ENV_PARENT_DIR - If you have multiple projects (all under then same directory) set this to the parent directory. If you only want to serve one project you should leave this empty and set the next variable.
• PROJECT_ENV - Set this to the project environment directory of you project if you serve only one. This variable is ignored if ENV_PARENT_DIR is set to anything but empty.
• PORT - This simply tells tracd on what port to listen.
• ADDITIONAL_OPTS - This variable allows you to pass additional parameters to the tracd. It's the place to add you authentication flags. If you have only a single project and would like to omit the project name in the url (e.g. something like http://trac.example.com) add the -s flag to this variable.

The other part of the configuration allows you to easily adjust the script to your machine. It has variables for specifying the path to the tracd executable and the paths of other things required of the correct operation of the script. In most cases you wouldn't have to change this part.

Now copy the script to /etc/init.d/trac and adjust the configuration to your needs. To enable the script on startup you should use your distro specific tools for managing the rc scripts. On fedora that means you will have to do chkconfig trac on. On Ubuntu you should do update-rc.d trac defaults.

I hope you will find the script useful. If you have any comments or suggestions regarding the script please comment.

UPDATE (13/10/2008):
Here is an version of the script for download.
UPDATE (17/2/2009):
See Starting tracd without Root Privileges at Startup for a different way to start tracd that doesn't require root privileges.