[BackupPC-users] Exceptionally slow rsync backup

Hi all,

I've got an unusual problem (well, unusual for me)...

I've had a backuppc install for many years, and it *never* deletes or
expires a backup. Every full and incremental backup ever taken is kept.
There are only about 5 old hosts, and the host with the most backups has
1705 backups.

Details from the webpage:

  * Uncompressed pool:
      o Pool is 357.46GB comprising 960280 files and 4369 directories
(as of 2012-04-14 01:05),
      o Pool hashing gives 152 repeated files with longest chain 7,
      o Nightly cleanup removed 0 files of size 0.00GB (around
2012-04-14 01:05),
  * Compressed pool:
      o Pool is 682.99GB comprising 3399853 files and 4369 directories
(as of 2012-04-14 01:34),
      o Pool hashing gives 1553 repeated files with longest chain 102,
      o Nightly cleanup removed 1093 files of size 0.01GB (around
2012-04-14 01:34),
  * Pool file system was recently at 20% (2012-04-14 13:14), today's max
is 20% (2012-04-14 01:00) and yesterday's max was 20%.

Now, all the old machines have been retired, and replacement machines
setup/configured with new names/etc.
However, all of the data has been retained, but moved into a different
folder structure (eg, what was in /home/username is now is
/home/archive/username).
Thus, the majority of the first backup is files which did not exist in
the previous backup (because it is the first backup) but already exist
in the pool.

Backups are exceptionally slow. The backup has been running for over 24
hours, and has backed up 23776 files in a directory of approx 170000
files. While this was in progress, I've used rsync -av --progress to
copy the same folder to another machine, and it copied over 73166 files
in approx 30 minutes (which is 20419M of data).

All machines are connected with gigabit ethernet.

So the source machine is certainly capable of backing up very quickly.

The backup machine (while the backup is still in progress):
hdparm -tT /dev/md2
/dev/md2:
 Timing cached reads:   1650 MB in  2.00 seconds = 825.39 MB/sec
 Timing buffered disk reads: 736 MB in  3.01 seconds = 244.52 MB/sec

This is a RAID6 array, using 2TB WD Caviar Black SATAII drives.
The OS drive is on a separate RAID1, on different physical drives.
/dev/md2 is mounted at /var/lib/backuppc

Again, while the backup is still running in the background, I did a
rsync -av --progress from the same source directory to
/var/lib/backuppc/trash which copied 10,000 files (2508M) in less than 5
minutes. Again, I don't think it is ethernet bandwidth, disk
performance, or CPU/memory (both machines are using zero swap).

Can anyone suggest a method to speed up the backups, or something that
can be looked at to see why this backup is taking so long?

It happens with more than one backup client, I've just used this one as
an example.

Regards,
Adam

-- 
Adam Goryachev
Website Managers
www.websitemanagers.com.au


------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/