Bacula 1.32a released
Kern Sibbald <[email protected]> 04 Oct 2003 18:27:59 +0200
| Newsgroups | gmane.comp.sysutils.backup.bacula.announce,gmane.comp.bacula.user |
|---|---|
| Message-ID | <1065284879.1198.621.camel@rufus> |
Hello, Until now, now Bacula bug has resulted in bad data being written to the Volume. I recently found the first such bug, and fortunately it is extremely rare, and only occurs if you are using the "sparse=yes" option on your Include, and if the first 1016 bytes of a 32K block are zero, and the last 248 bytes of the block are zero, then Bacula 1.32 and lower will incorrectly assume the whole block is zero and thus need not be backed up. Version 1.32a fixes this bug, and also fixes "list nextvol" so that it respects pool overrides in the run records. You only need to upgrade if you are using the sparse option or you count on "list nextvol" working correctly. If you are using the sparse option, I strongly recommend that you upgrade and do a Full backup, especially if you are using FreeBSD where the strange pattern of zeros apparently occurs more often. At the same time, I have released new Windows binaries. Well, I wonder if I will sleep better at night knowing that I killed off the one and hopefully only serious bug. Best regards, Kern ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf