Revision 8676899f
Added by Phil Davis over 10 years ago
| etc/rc.backup_rrd.sh | ||
|---|---|---|
| 7 | 7 |
if [ -d "${RRDDBPATH}" ]; then
|
| 8 | 8 |
[ -z "$NO_REMOUNT" ] && /etc/rc.conf_mount_rw |
| 9 | 9 |
[ -f "${CF_CONF_PATH}/rrd.tgz" ] && /bin/rm -f "${CF_CONF_PATH}"/rrd.tgz
|
| 10 |
|
|
| 10 |
|
|
| 11 |
tgzlist="" |
|
| 12 |
|
|
| 11 | 13 |
for rrdfile in "${RRDDBPATH}"/*.rrd ; do
|
| 12 | 14 |
xmlfile="${rrdfile%.rrd}.xml"
|
| 15 |
tgzfile="${rrdfile%.rrd}.tgz"
|
|
| 13 | 16 |
/usr/bin/nice -n20 /usr/local/bin/rrdtool dump "$rrdfile" "$xmlfile" |
| 14 |
cd / && /usr/bin/tar -rf "${CF_CONF_PATH}"/rrd.tar -C / "${RRDDBPATH#/}"/*.xml
|
|
| 17 |
cd / && /usr/bin/tar -czf "${tgzfile}" -C / "${RRDDBPATH#/}"/*.xml
|
|
| 15 | 18 |
/bin/rm -f "${RRDDBPATH}"/*.xml
|
| 19 |
tgzlist="${tgzlist} @${tgzfile}"
|
|
| 16 | 20 |
done |
| 17 |
if [ -f "${CF_CONF_PATH}/rrd.tar" ]; then
|
|
| 18 |
/usr/bin/gzip "${CF_CONF_PATH}/rrd.tar"
|
|
| 19 |
/bin/mv "${CF_CONF_PATH}/rrd.tar.gz" "${CF_CONF_PATH}/rrd.tgz"
|
|
| 21 |
if [ -n "${tgzlist}" ]; then
|
|
| 22 |
cd / && /usr/bin/tar -czf "${CF_CONF_PATH}/rrd.tgz" ${tgzlist}
|
|
| 23 |
/bin/rm -f "${RRDDBPATH}"/*.tgz
|
|
| 20 | 24 |
fi |
| 21 | 25 |
[ -z "$NO_REMOUNT" ] && /etc/rc.conf_mount_ro |
| 22 | 26 |
fi |
Also available in: Unified diff
Process RRD backup compression in var
Prior to this the RRD xml files were added uncompressed to the archive in /cf/conf and then that archive was compressed at the end.
My /cf partition is only 50MB. The uncompressed archive of all the xml files is already 35MB. With a few config backups, or a few more VLANs (xml files) I will soon run into the 50MB limit.
This change creates each xml from rrd one at a time, then compresses that 1 xml into a tgz in /var/db/rrd, deletes the xml then loops to the next rrd file.
At the end of the loop, there are a bunch of /var/db/rrd/*.tgz files, which are small (they take up <2MB on my system, from XMLs that total 35MB). They are then unpacked and put into 1 /cf/conf/rrd.tgz in a single command.
Thus there is no time when 35MB of xml content has to be stored anywhere.
This should work for systems with a lot of RRD files that turn into XML and then TGZ one at a time.