[Box Backup] exception BackupStore InvalidBackupStoreFilename (4/3)

Jason boxbackup@fluffy.co.uk
Mon, 5 May 2008 18:38:59 -0400


# time cat 1_gig_file > /dev/null
real    0m58.625s
user    0m0.100s
sys     0m6.720s

I definitely think it's CPU bound because of the drive encryption.
One of the CPUs is pegged by kernel processes when I run that cat.
Unfortunately I don't have enough unencrypted free space on that box
to copy the data there.  How would I go about copying the data to
another system and fixing it there?  Do I just need to copy this one
store and the boxbackup config files to the other machine?

I store it on an encrypted filesystem because I already have a cluster
of boxes dedicated to data storage and the whole thing is encrypted,
so it was far easier for me to just put the boxbackup stores with all
my other data instead of sectioning of an unexcrypted bit just for
boxbackup.

Thank you very much for your continued help with this.  It really
bolsters my confidence in boxbackup knowing that someone actually
cares.

Jason


On Mon, May 5, 2008 at 12:26 PM, Chris Wilson <chris@qwirx.com> wrote:
> Hi James,
>
>
>  On Mon, 5 May 2008, Chris Wilson wrote:
>
>  > Out of curiosity, roughly how many "object is unattached" messages did
>  > you get in the check run (without fix), and how long is it taking to
>  > produce each one when fixing? (this should give us an estimate of how
>  > long the fix will take at the current rate).
>
>  Sorry, I missed that part in your previous message where you answered one
>  of these. If every single object in that range is missing a parent then
>  it's about 120,000 objects, of whicit's finished fixing about 5,000 so
>  far, in just over a day. That tells me that it's taking about 16 seconds
>  each, and it's about 5% finished, so it will take 20-30 days to finish :-(
>
>  I think the main problem is this code in BackupStoreCheck2.cpp:
>
>         // The easiest way to do this is to verify it again. Not such a
>         // bad penalty, because this really shouldn't be done very often.
>
>         std::auto_ptr<RaidFileRead> file(RaidFileRead::Open(mDiscSetNumber,
>                 filename));
>         BackupStoreFile::VerifyEncodedFileFormat(*file, &diffFromObjectID);
>
>  Unfortunately this is done for every unattached file, in your case
>  120,000 of them. That's not compatible with my definition of "not very
>  often".
>
>  I think it means reading the entire contents of the file in each case. I
>  don't know how many files you have in total, or what the encrypted
>  filesystem is that you're running on, but I guess it's conceiveable that
>  it could take days to read 20 GB of data, especially if seeks are
>  involved.
>
>  What sort of performance do you get, e.g. how long does it take to read a
>  1GB file from this filesystem? Is it CPU or I/O bound? If it's CPU, then
>  you might find it much faster to copy the data off the encrypted partition
>  before running check fix on it.
>
>  Also, since Box Backup's backups are already encrypted by the client, why
>  put it on an encrypted filesystem?
>
>
>
>  Cheers, Chris.
>  --
>  _____ __     _
>  \  __/ / ,__(_)_  | Chris Wilson <0000 at qwirx.com> - Cambs UK |
>  / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL Developer |
>  \ _/_/_/_//_/___/ | We are GNU : free your mind & your software |
>  _______________________________________________
>  boxbackup mailing list
>  boxbackup@fluffy.co.uk
>  http://lists.warhead.org.uk/mailman/listinfo/boxbackup
>