On Tue, Feb 26, 2008 at 2:23 PM, Tomasz Chmielewski <[EMAIL PROTECTED]> wrote: > > Can you give us more details on your disk array? Controller, disks, > > RAID layout, ext3 fs creation options, etc... > > I said some of that already - but here are some missing parts. > 5x 400 MB HDD (WDC WD4000YR) > The controller uses sata_mv module (Marvell). > > Linux software RAID over it, LVM over it. Filesystem for backup is 1.2 > TB big. > The fs has these features: has_journal filetype needs_recovery > sparse_super large_file.
Let me see if I can summarize this properly: iSCSI to a Thecus n5200 with 5x 400 MB HDD (WDC WD4000YR). The box runs a custom Debian kernel. The box uses software raid 5 + LVM. You didn't mention stripe size of the raid 5. You also didn't mention whether you used the stride option when creating the ext3 filesystem. > On an empty filesystem on than NAS I can write with ~25 MB/s. > I guess that backup filesystem is just very fragmented? > On the other hand, it is only 60-70% full, so drop from ~25 MB/s (empty > filesystem) to ~1.3 MB/s is indeed something odd. 25 MB/s seems abysmal for an empty filesystem. I really would expect at least twice that on a GigE iSCSI network, and probably closer to 75 MB/s. > But hey, sequential writing isn't something very often used with BackupPC. > Just how fast can you write a file downloaded from the internet (or even > LAN) which is being compressed with bzip2 at the same time. Or better > yet, when there are 10 such threads compressing with bzip2. No, sequential write performance isn't used much for BackupPC, but low numbers do indicate that there is some sort of bottleneck there. FWIW, on my BackupPC system here with a simple software RAID1 of two 250MB ATA disks can write 25MB/s sequentially on it's BackupPC partition which is 56% full. > Would BackupPC detect a corrupted file which is unique? Even if yes, it > would be corrupted, with no other copy, so... BackupPC does detect when files on disk get corrupted when new backups are being made (since it compares files being backed up directly against the files in the pool). They are checked less frequently when the rsync checksum-seed option is used, but you can change that. See the docs for more info. Anyway, looking at your setup, you have a number of things which are contributing to your performance. * Software RAID5 - RAID5 is HORRIBLE (yes, it needs to be capitalized!) for small reads and writes. Especially for small random writes, performance can degrade to speeds well below the speed of a single disk. Small random reads will also perform poorly and at best will match the speed of a single disk. Hardware RAID5 with a large enough battery backed cache can negate some of the performance drawbacks of small random reads/writes * LVM - Often LVM can slow things down as well since it adds another layer of abstraction. * iScsi - Network based filesystem is never going to be as fast as directly attached storage. * Unlikely that the proper ext3 stride option was used to create the filesystem which can result in poor performance on a striped RAID array. I'm not sure how LVM would further affect this. * Small amount of memory on NAS - It appears that your Thecus 5200 only has 256MB of RAM. I would expect that having significantly more memory could help IO performance. So there you go. IMO, unless you are willing to overhaul your storage system or slightly increase the risk of data corruption (IMO, data=writeback instead of the default data=ordered should be a large gain for you and is very safe), you are going to continue to fall further behind in your nightly cleanup runs. -Dave ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ BackupPC-users mailing list [email protected] List: https://lists.sourceforge.net/lists/listinfo/backuppc-users Wiki: http://backuppc.wiki.sourceforge.net Project: http://backuppc.sourceforge.net/
