Digest and Inventory files

robb robb at canfield.com
Sat Aug 25 16:43:28 UTC 2007


During testing I noticed that the digest database file is named per the
SOURCE settings (default .brackup-digest.db). This causes a problem when
doing geo-reduandnt backups. Since the digest is used to see if a file
has been backed up it skips files that may be one one but not the other
of multiple TARGETS.

Of course multiple SOURCE sections can be created, but that becomes a
bit of a maintenance problem.

Wouldn't it be better to use a naming convention like that used for the
INVENTORY database file? That is one that includes the target name. If
you WANT the old behavior then specify the digest file name in the TARGET.

The other issue I encountered was that the Inventory database is not
recovered during a restore and is not backed up in an ATOMIC fashion.
This means that after a recovery the next backup will re-encrypt and
upload EVERY file. Has anyone else seen this?

So far SFTP testing is going well. I have backed up and restored 700+
MB! YEA! I also added onerror handling to restore AND backup. It's
possible that during backup a file goes missing (temporary files, pid
files, etc. that may not be in the ignore list). This will halt the
backup which is not good when the backup is fully automated. By
specifying --onerror=continue (or prompt) the code will skip past the
missing file and continue. I am trusting garbage collection to remove
any blocks at a later time.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3237 bytes
Desc: S/MIME Cryptographic Signature
Url : http://lists.danga.com/pipermail/brackup/attachments/20070825/cd351113/smime.bin


More information about the brackup mailing list