ccollect/ccollect.sh

610 lines
14 KiB
Bash
Raw Normal View History

2005-11-17 11:33:49 +00:00
#!/bin/sh
2008-03-07 22:03:55 +00:00
#
# 2005-2008 Nico Schottelius (nico-ccollect at schottelius.org)
#
# This file is part of ccollect.
#
# ccollect is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# ccollect is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with ccollect. If not, see <http://www.gnu.org/licenses/>.
#
2007-08-14 12:33:22 +00:00
# Initially written for SyGroup (www.sygroup.ch)
2005-11-17 11:33:49 +00:00
# Date: Mon Nov 14 11:45:11 CET 2005
#
2005-12-06 14:55:07 +00:00
# where to find our configuration and temporary file
#
CCOLLECT_CONF=${CCOLLECT_CONF:-/etc/ccollect}
2007-08-16 18:17:54 +00:00
CSOURCES=${CCOLLECT_CONF}/sources
CDEFAULTS=${CCOLLECT_CONF}/defaults
CPREEXEC="${CDEFAULTS}/pre_exec"
CPOSTEXEC="${CDEFAULTS}/post_exec"
2007-08-16 18:17:54 +00:00
TMP=$(mktemp "/tmp/$(basename $0).XXXXXX")
VERSION=0.7.1
RELEASE="2008-04-XX"
2007-08-16 18:17:54 +00:00
HALF_VERSION="ccollect ${VERSION}"
FULL_VERSION="ccollect ${VERSION} (${RELEASE})"
#
# CDATE: how we use it for naming of the archives
# DDATE: how the user should see it in our output (DISPLAY)
#
2007-08-16 18:17:54 +00:00
CDATE="date +%Y%m%d-%H%M"
2007-08-14 12:33:22 +00:00
DDATE="date +%Y-%m-%d-%H:%M:%S"
#
# unset parallel execution
#
PARALLEL=""
2005-12-06 14:55:07 +00:00
#
# catch signals
#
2007-08-16 18:17:54 +00:00
trap "rm -f \"${TMP}\"" 1 2 15
2005-11-17 11:33:49 +00:00
2007-01-20 15:41:38 +00:00
#
# Functions
#
# time displaying echo
_techo()
{
2007-08-16 18:31:30 +00:00
echo "$(${DDATE}): $@"
}
# exit on error
2007-01-20 15:41:38 +00:00
_exit_err()
{
_techo "$@"
2007-08-16 18:17:54 +00:00
rm -f "${TMP}"
2007-01-20 15:41:38 +00:00
exit 1
}
add_name()
{
2007-08-16 18:17:54 +00:00
sed "s:^:\[${name}\] :"
}
2008-03-14 05:50:24 +00:00
pcmd()
{
if [ "$remote_host" ]; then
ssh "$remote_host" "$@"
2008-03-14 05:50:24 +00:00
else
"$@"
fi
}
2007-08-17 22:14:10 +00:00
#
# Version
#
display_version()
{
echo "${FULL_VERSION}"
exit 0
}
#
2005-12-06 12:45:37 +00:00
# Tell how to use us
#
2005-12-06 12:45:37 +00:00
usage()
{
echo "$(basename $0): <interval name> [args] <sources to backup>"
2005-12-06 12:45:37 +00:00
echo ""
2006-01-24 21:36:28 +00:00
echo " ccollect creates (pseudo) incremental backups"
2005-12-06 12:45:37 +00:00
echo ""
echo " -h, --help: Show this help screen"
2006-01-24 21:58:39 +00:00
echo " -p, --parallel: Parallelise backup processes"
echo " -a, --all: Backup all sources specified in ${CSOURCES}"
2007-03-13 07:47:04 +00:00
echo " -v, --verbose: Be very verbose (uses set -x)"
2007-08-17 22:14:10 +00:00
echo " -V, --version: Print version information"
2005-12-06 12:45:37 +00:00
echo ""
echo " This is version ${VERSION}, released on ${RELEASE}"
echo " (the first version was written on 2005-12-05 by Nico Schottelius)."
2006-01-24 21:58:39 +00:00
echo ""
2006-10-15 19:56:55 +00:00
echo " Retrieve latest ccollect at http://unix.schottelius.org/ccollect/"
2005-12-06 12:45:37 +00:00
exit 0
}
#
# need at least interval and one source or --all
#
if [ $# -lt 2 ]; then
2007-08-17 22:14:10 +00:00
if [ "$1" = "-V" -o "$1" = "--version" ]; then
display_version
else
usage
fi
fi
#
# check for configuraton directory
#
[ -d "${CCOLLECT_CONF}" ] || _exit_err "No configuration found in " \
2007-08-16 18:17:54 +00:00
"\"${CCOLLECT_CONF}\" (is \$CCOLLECT_CONF properly set?)"
2005-12-06 12:45:37 +00:00
#
# Filter arguments
#
2007-05-31 09:52:33 +00:00
export INTERVAL="$1"; shift
2005-12-06 12:45:37 +00:00
i=1
2006-04-29 10:58:57 +00:00
no_sources=0
2005-12-06 12:45:37 +00:00
#
# Create source "array"
#
while [ "$#" -ge 1 ]; do
2007-08-14 13:30:00 +00:00
eval arg=\"\$1\"; shift
2006-10-25 12:48:48 +00:00
2007-08-16 18:17:54 +00:00
if [ "${NO_MORE_ARGS}" = 1 ]; then
eval source_${no_sources}=\"${arg}\"
no_sources=$((${no_sources}+1))
2007-05-31 09:55:37 +00:00
# make variable available for subscripts
eval export source_${no_sources}
else
2007-08-16 18:17:54 +00:00
case "${arg}" in
-a|--all)
ALL=1
;;
2005-12-08 15:14:46 +00:00
-v|--verbose)
VERBOSE=1
;;
-p|--parallel)
PARALLEL=1
;;
-h|--help)
usage
;;
--)
NO_MORE_ARGS=1
;;
*)
2006-04-29 10:58:57 +00:00
eval source_${no_sources}=\"$arg\"
no_sources=$(($no_sources+1))
;;
esac
fi
2005-12-06 12:45:37 +00:00
i=$(($i+1))
2005-12-06 12:45:37 +00:00
done
2007-05-31 09:55:37 +00:00
# also export number of sources
export no_sources
2005-12-08 15:14:46 +00:00
#
# be really, really, really verbose
2005-12-08 15:14:46 +00:00
#
2007-08-16 18:17:54 +00:00
if [ "${VERBOSE}" = 1 ]; then
2005-12-08 15:14:46 +00:00
set -x
fi
2005-12-06 14:35:29 +00:00
#
# Look, if we should take ALL sources
#
2007-08-16 18:17:54 +00:00
if [ "${ALL}" = 1 ]; then
2005-12-06 14:35:29 +00:00
# reset everything specified before
2006-04-29 10:58:57 +00:00
no_sources=0
2006-10-25 12:48:48 +00:00
2005-12-06 14:55:07 +00:00
#
# get entries from sources
#
cwd=$(pwd -P)
2007-08-14 13:30:00 +00:00
( cd "${CSOURCES}" && ls > "${TMP}" ); ret=$?
2006-10-15 14:50:12 +00:00
2007-08-14 13:30:00 +00:00
[ "${ret}" -eq 0 ] || _exit_err "Listing of sources failed. Aborting."
2006-10-25 12:48:48 +00:00
2005-12-06 14:55:07 +00:00
while read tmp; do
2007-08-16 18:17:54 +00:00
eval source_${no_sources}=\"${tmp}\"
no_sources=$((${no_sources}+1))
done < "${TMP}"
2005-12-06 14:35:29 +00:00
fi
#
# Need at least ONE source to backup
#
if [ "${no_sources}" -lt 1 ]; then
usage
else
_techo "${HALF_VERSION}: Beginning backup using interval ${INTERVAL}"
fi
#
# Look for pre-exec command (general)
#
if [ -x "${CPREEXEC}" ]; then
2007-08-16 18:17:54 +00:00
_techo "Executing ${CPREEXEC} ..."
"${CPREEXEC}"; ret=$?
2007-08-16 19:37:39 +00:00
_techo "Finished ${CPREEXEC} (return code: ${ret})."
2007-08-16 19:37:39 +00:00
[ "${ret}" -eq 0 ] || _exit_err "${CPREEXEC} failed. Aborting"
fi
#
# check default configuration
#
D_FILE_INTERVAL="${CDEFAULTS}/intervals/${INTERVAL}"
D_INTERVAL=$(cat "${D_FILE_INTERVAL}" 2>/dev/null)
#
# Let's do the backup
#
i=0
2007-08-16 18:17:54 +00:00
while [ "${i}" -lt "${no_sources}" ]; do
2005-11-17 11:33:49 +00:00
#
2006-04-29 10:58:57 +00:00
# Get current source
#
2006-11-14 23:01:19 +00:00
eval name=\"\$source_${i}\"
2007-08-16 18:17:54 +00:00
i=$((${i}+1))
export name
#
# start ourself, if we want parallel execution
#
if [ "${PARALLEL}" ]; then
"$0" "${INTERVAL}" "${name}" &
continue
fi
#
# Start subshell for easy log editing
#
(
#
# Stderr to stdout, so we can produce nice logs
#
exec 2>&1
#
# Configuration
#
backup="${CSOURCES}/${name}"
c_source="${backup}/source"
c_dest="${backup}/destination"
c_exclude="${backup}/exclude"
c_verbose="${backup}/verbose"
c_vverbose="${backup}/very_verbose"
c_rsync_extra="${backup}/rsync_options"
c_summary="${backup}/summary"
c_pre_exec="${backup}/pre_exec"
c_post_exec="${backup}/post_exec"
f_incomplete="delete_incomplete"
c_incomplete="${backup}/${f_incomplete}"
2008-03-14 05:50:24 +00:00
c_remote_host="${backup}/remote_host"
2007-08-16 18:31:30 +00:00
#
# Marking backups: If we abort it's not removed => Backup is broken
#
c_marker=".ccollect-marker"
2006-10-26 08:30:14 +00:00
2007-01-20 15:41:38 +00:00
#
# Times
2007-01-20 15:41:38 +00:00
#
begin_s=$(date +%s)
#
# unset possible options
#
EXCLUDE=""
RSYNC_EXTRA=""
SUMMARY=""
VERBOSE=""
VVERBOSE=""
2007-08-16 18:31:30 +00:00
DELETE_INCOMPLETE=""
_techo "Beginning to backup"
#
# Standard configuration checks
#
2007-03-13 07:47:04 +00:00
if [ ! -e "${backup}" ]; then
2007-08-16 18:17:54 +00:00
_exit_err "Source does not exist."
fi
2006-10-25 12:48:48 +00:00
#
# configuration _must_ be a directory
#
2007-03-13 07:47:04 +00:00
if [ ! -d "${backup}" ]; then
2007-08-16 18:17:54 +00:00
_exit_err "\"${name}\" is not a cconfig-directory. Skipping."
fi
#
# first execute pre_exec, which may generate destination or other
# parameters
#
2007-03-13 07:47:04 +00:00
if [ -x "${c_pre_exec}" ]; then
_techo "Executing ${c_pre_exec} ..."
2007-03-13 07:47:04 +00:00
"${c_pre_exec}"; ret="$?"
_techo "Finished ${c_pre_exec} (return code ${ret})."
2007-03-13 07:47:04 +00:00
if [ "${ret}" -ne 0 ]; then
_exit_err "${c_pre_exec} failed. Skipping."
fi
fi
#
# interval definition: First try source specific, fallback to default
#
2007-03-13 07:47:04 +00:00
c_interval="$(cat "${backup}/intervals/${INTERVAL}" 2>/dev/null)"
2007-03-13 07:47:04 +00:00
if [ -z "${c_interval}" ]; then
c_interval="${D_INTERVAL}"
2007-03-13 07:47:04 +00:00
if [ -z "${c_interval}" ]; then
_exit_err "No definition for interval \"${INTERVAL}\" found. Skipping."
fi
fi
#
# Source checks
#
2007-03-13 07:47:04 +00:00
if [ ! -f "${c_source}" ]; then
_exit_err "Source description \"${c_source}\" is not a file. Skipping."
else
2008-03-14 05:50:24 +00:00
source=$(cat "${c_source}"); ret="$?"
if [ "${ret}" -ne 0 ]; then
_exit_err "Source ${c_source} is not readable. Skipping."
fi
fi
#
2008-03-14 05:50:24 +00:00
# Destination is a path
#
2008-03-04 23:39:17 +00:00
if [ ! -f "${c_dest}" ]; then
_exit_err "Destination ${c_dest} is not a file. Skipping."
else
2008-03-14 05:50:24 +00:00
ddir=$(cat "${c_dest}"); ret="$?"
2008-03-04 23:39:17 +00:00
if [ "${ret}" -ne 0 ]; then
_exit_err "Destination ${c_dest} is not readable. Skipping."
fi
fi
2008-03-14 05:50:24 +00:00
#
# do we backup to a remote host? then set pre-cmd
#
if [ -f "${c_remote_host}" ]; then
# adjust ls and co
remote_host=$(cat "${c_remote_host}"); ret="$?"
if [ "${ret}" -ne 0 ]; then
_exit_err "Remote host file ${c_remote_host} exists, but is not readable. Skipping."
fi
destination="${remote_host}:${ddir}"
else
remote_host=""
2008-03-14 05:50:24 +00:00
destination="${ddir}"
fi
export remote_host
2008-03-14 05:50:24 +00:00
#
# check for existence / use real name
#
pcmd cd "$ddir" || _exit_err "Cannot change to ${ddir}. Skipping."
2008-03-14 05:50:24 +00:00
2007-08-16 18:31:30 +00:00
#
# Check whether to delete incomplete backups
#
if [ -f "${c_incomplete}"
-o -f "${CDEFAULTS}/${f_incomplete}]; then
2007-08-16 18:31:30 +00:00
DELETE_INCOMPLETE="yes"
fi
2007-08-16 18:08:23 +00:00
# NEW method as of 0.6:
# - insert ccollect default parameters
# - insert options
# - insert user options
2007-08-14 13:30:00 +00:00
2007-08-14 13:40:23 +00:00
#
# rsync standard options
#
2007-08-16 18:08:23 +00:00
set -- "$@" "--archive" "--delete" "--numeric-ids" "--relative" \
2008-03-04 23:39:17 +00:00
"--delete-excluded" "--sparse"
2007-08-16 18:08:23 +00:00
#
# exclude list
#
if [ -f "${c_exclude}" ]; then
2007-08-16 18:17:54 +00:00
set -- "$@" "--exclude-from=${c_exclude}"
fi
2006-10-25 12:48:48 +00:00
#
2006-01-24 20:46:49 +00:00
# Output a summary
#
2007-03-13 07:47:04 +00:00
if [ -f "${c_summary}" ]; then
2007-08-14 13:40:23 +00:00
set -- "$@" "--stats"
2006-01-24 20:46:49 +00:00
fi
2006-10-25 12:48:48 +00:00
#
# Verbosity for rsync
#
2007-08-14 13:40:23 +00:00
if [ -f "${c_vverbose}" ]; then
set -- "$@" "-vv"
elif [ -f "${c_verbose}" ]; then
set -- "$@" "-v"
2006-03-19 00:10:58 +00:00
fi
2006-10-25 12:48:48 +00:00
#
2007-08-14 13:40:23 +00:00
# extra options for rsync provided by the user
#
2007-08-14 13:40:23 +00:00
if [ -f "${c_rsync_extra}" ]; then
while read line; do
set -- "$@" "$line"
done < "${c_rsync_extra}"
fi
2006-10-25 12:48:48 +00:00
2007-08-16 18:08:23 +00:00
#
# Check for incomplete backups
#
pcmd ls -1 "$ddir/${INTERVAL}"*".${c_marker}" > "${TMP}" 2>/dev/null
2007-08-16 19:25:46 +00:00
i=0
2007-08-16 19:25:46 +00:00
while read incomplete; do
2008-03-17 11:06:17 +00:00
eval incomplete_$i=\"$(echo ${incomplete} | sed "s/\\.${c_marker}\$//")\"
i=$(($i+1))
done < "${TMP}"
2008-03-17 11:06:17 +00:00
j=0
while [ "$j" -lt "$i" ]; do
2008-03-17 11:11:09 +00:00
eval realincomplete=\"\$incomplete_$j\"
2007-08-16 19:25:46 +00:00
_techo "Incomplete backup: ${realincomplete}"
if [ "${DELETE_INCOMPLETE}" = "yes" ]; then
_techo "Deleting ${realincomplete} ..."
pcmd rm $VVERBOSE -rf "${ddir}/${realincomplete}" || \
2007-08-27 17:00:43 +00:00
_exit_err "Removing ${realincomplete} failed."
2007-08-16 19:25:46 +00:00
fi
2008-03-17 11:08:40 +00:00
j=$((j+1))
done
#
# check if maximum number of backups is reached, if so remove
2007-08-16 18:41:03 +00:00
# use grep and ls -p so we only look at directories
#
2008-03-17 11:08:40 +00:00
count="$(pcmd ls -p1 "${ddir}" | grep "^${INTERVAL}\..*/\$" | wc -l \
| sed 's/^ *//g')" || _exit_err "Counting backups failed"
2007-08-16 18:41:03 +00:00
_techo "Existing backups: ${count} Total keeping backups: ${c_interval}"
if [ "${count}" -ge "${c_interval}" ]; then
substract=$((${c_interval} - 1))
remove=$((${count} - ${substract}))
2007-08-16 19:26:56 +00:00
_techo "Removing ${remove} backup(s)..."
2007-08-16 18:41:03 +00:00
pcmd ls -p1 "$ddir" | grep "^${INTERVAL}\..*/\$" | \
2008-03-17 11:09:30 +00:00
sort -n | head -n "${remove}" > "${TMP}" || \
_exit_err "Listing old backups failed"
2007-08-16 19:37:39 +00:00
i=0
2007-08-16 19:37:39 +00:00
while read to_remove; do
eval remove_$i=\"${to_remove}\"
i=$(($i+1))
done < "${TMP}"
2008-03-17 11:08:40 +00:00
j=0
2008-03-17 11:11:09 +00:00
while [ "$j" -lt "$i" ]; do
2008-03-17 11:21:35 +00:00
eval to_remove=\"\$remove_$j\"
2007-08-16 19:37:39 +00:00
_techo "Removing ${to_remove} ..."
2008-03-17 11:21:35 +00:00
pcmd rm ${VVERBOSE} -rf "${ddir}/${to_remove}" || \
2007-08-16 19:37:39 +00:00
_exit_err "Removing ${to_remove} failed."
2008-03-17 11:11:09 +00:00
j=$((j+1))
done
2007-08-16 18:41:03 +00:00
fi
2007-08-16 18:08:23 +00:00
#
2007-08-17 21:31:45 +00:00
# Check for backup directory to clone from: Always clone from the latest one!
#
# Use ls -1c instead of -1t, because last modification maybe the same on all
# and metadate update (-c) is updated by rsync locally.
2007-08-16 18:08:23 +00:00
#
last_dir="$(pcmd ls -tcp1 "${ddir}" | grep '/$' | head -n 1)" || \
2008-03-14 05:50:24 +00:00
_exit_err "Failed to list contents of ${ddir}."
#
2007-08-16 19:37:39 +00:00
# clone from old backup, if existing
#
if [ "${last_dir}" ]; then
2008-03-17 10:44:20 +00:00
set -- "$@" "--link-dest=${ddir}/${last_dir}"
_techo "Hard linking from ${last_dir}"
fi
# set time when we really begin to backup, not when we began to remove above
destination_date=$(${CDATE})
destination_dir="${ddir}/${INTERVAL}.${destination_date}.$$"
destination_full="${destination}/${INTERVAL}.${destination_date}.$$"
2006-10-25 12:48:48 +00:00
# give some info
_techo "Beginning to backup, this may take some time..."
2007-08-16 19:45:02 +00:00
_techo "Creating ${destination_dir} ..."
pcmd mkdir ${VVERBOSE} "${destination_dir}" || \
_exit_err "Creating ${destination_dir} failed. Skipping."
#
# added marking in 0.6 (and remove it, if successful later)
#
pcmd touch "${destination_dir}.${c_marker}"
2005-12-06 16:08:38 +00:00
#
# the rsync part
#
2006-10-25 12:48:48 +00:00
2007-08-14 13:40:23 +00:00
_techo "Transferring files..."
rsync "$@" "${source}" "${destination_full}"; ret=$?
2006-10-25 13:16:37 +00:00
#
2007-08-16 18:08:23 +00:00
# remove marking here
2006-10-25 13:16:37 +00:00
#
pcmd rm "${destination_dir}.${c_marker}" || \
_exit_err "Removing ${destination_dir}/${c_marker} failed."
2007-08-16 18:08:23 +00:00
_techo "Finished backup (rsync return code: $ret)."
2007-08-20 15:33:09 +00:00
if [ "${ret}" -ne 0 ]; then
_techo "Warning: rsync exited non-zero, the backup may be broken (see rsync errors)."
fi
#
# post_exec
#
2007-08-16 18:08:23 +00:00
if [ -x "${c_post_exec}" ]; then
_techo "Executing ${c_post_exec} ..."
"${c_post_exec}"; ret=$?
_techo "Finished ${c_post_exec}."
2006-03-19 00:13:38 +00:00
2007-08-16 18:08:23 +00:00
if [ ${ret} -ne 0 ]; then
_exit_err "${c_post_exec} failed."
2006-03-19 00:13:38 +00:00
fi
fi
2007-08-16 18:08:23 +00:00
# Calculation
end_s=$(date +%s)
2006-10-15 19:01:28 +00:00
full_seconds=$((${end_s} - ${begin_s}))
hours=$((${full_seconds} / 3600))
seconds=$((${full_seconds} - (${hours} * 3600)))
minutes=$((${seconds} / 60))
2006-10-15 19:01:28 +00:00
seconds=$((${seconds} - (${minutes} * 60)))
2007-08-16 18:08:23 +00:00
_techo "Backup lasted: ${hours}:${minutes}:${seconds} (h:m:s)"
) | add_name
done
2005-12-06 14:55:07 +00:00
2005-12-06 16:08:38 +00:00
#
# Be a good parent and wait for our children, if they are running wild parallel
#
2007-08-16 18:08:23 +00:00
if [ "${PARALLEL}" ]; then
2007-08-16 19:37:39 +00:00
_techo "Waiting for children to complete..."
2005-12-06 16:08:38 +00:00
wait
fi
#
# Look for post-exec command (general)
#
2007-08-16 18:08:23 +00:00
if [ -x "${CPOSTEXEC}" ]; then
_techo "Executing ${CPOSTEXEC} ..."
2007-08-16 19:37:39 +00:00
"${CPOSTEXEC}"; ret=$?
_techo "Finished ${CPOSTEXEC} (return code: ${ret})."
2006-10-25 12:48:48 +00:00
2007-08-16 18:08:23 +00:00
if [ ${ret} -ne 0 ]; then
2007-08-16 19:46:02 +00:00
_techo "${CPOSTEXEC} failed."
2006-03-19 00:13:38 +00:00
fi
fi
2007-08-16 18:08:23 +00:00
rm -f "${TMP}"
_techo "Finished ${WE}"