#!/bin/sh
# Name: mybin/apache/from-iserver	Author: js-cgi@inwap.com 06-Feb-2002
# Purpose: Pull Apache logs from a Verio iServer
# Normally produces no output.  Use "sum-log -v" for verbose mode.
# Server access logs expected in www/logs/access_log and www/logs/access_log*gz
# Server error  logs expected in www/logs/error_log  and www/logs/error_log*gz

#   Programs used:	(see http://www.inwap.com/mybin/ for sources)
# mv-web-logs	Moves compressed log files, runs logresolve on access_log
# rm-same	Delete files on web server if MD5 checksums match
#
#   Files used:
# access_log access_log.?.gz	Apache logs (without DNS lookups)
# error_log   error_log.?.gz	Apache error log
#   $LOGDIR/yyyy/mmdd-acc.gz	Archived access logs (with names for IP addrs)

LOGDIR=$HOME/weblogs		# Where to put the results
BINDIR=$HOME/mybin		# Parent of dirs of other progs
PATH=$BINDIR/miscunix:$BINDIR/apache:$PATH

start=`date +%T`

if [ x$1 = x-v ]; then
  v=-v ; shift			# Use -v for verbose mode
fi

WEBSERVER=$1
if [ $WEBSERVER: = : ]; then
  cat <<EOM >&2
Usage: $0 [-v] www.iserver.com
  This script uses rsync and ssh to fetch www/logs from a Verio iServer.
  Use the -v option for verbose mode.
EOM
  exit 1;
fi

[ x$v = x-v ] && set -v
# Pull updated files from the web server
# Fetch ftp/anon/incoming, users/*/mail, www/logs/*log, www/htdocs/*.count
cd $HOME
rsync -e ssh -au $v $WEBSERVER:'{bin,www}' . | egrep -v '/$'

# Remove old log files after verifying checksums

rm-same $WEBSERVER www/logs/access_log.?.gz www/logs/error_log.?.gz

# Do DNS lookups via logresolve, save logs into a yearly subdirectory
start2=`date +%T`
mv-web-logs $v www/logs/access_log.?.gz www/logs/error_log.?.gz $LOGDIR
if [ x$v = x-v ]; then
  du $LOGDIR/20[0-9][0-9]	# Disk usage by year
  echo "Times: $start, $start2," `date +%T`
fi
# End of mybin/apache/from-iserver