Hello Bastian, Sparse file handling in Bacula was intended to work with sparse files, and in most cases also works with non-sparse files. You are applying it to raw devices.
When Bacula's sparse file handling skips blocks, and those blocks are re-read, the program gets zeros, so in 99.99% of the cases, the program cannot distinguish the difference. In your case, apparently holes in a VM and zeros are not the same, and thus sparse file handling is not appropriate. It was never intended to work with raw devices that are VM images. The solution is very simple: do not use sparse file handling in those cases. Best regards, Kern On 08/07/2012 08:24 AM, Bastian Friedrich wrote: > Am Montag, 6. August 2012, 20:35:52 schrieb Radosław Korzeniewski: >> If a user will have a thin-provisioned volume that he can successfully use >> sparse=yes option without any garbage and can use it even with a vm image. > I'm not sure whether I understand correctly what you are writing; however, I > suppose you are still at risk of data corruption. > > If you have a VM running on an LV, and you have a _file_ inside that VM that > contains large chunks of zeros, these zeros are indistinguishable from "empty" > blocks. Thus, during restore, the respective blocks are not written, > (possibly) resulting in garbage on disk -- and thus garbage in said file of > the VM. > > Try it. > > Bastian > ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Bacula-devel mailing list Bacula-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-devel
