Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2006-11-10 08:09:40 -0800 \ Carson Gaspar: > Lars Ellenberg wrote: > >say, some "invalid" data is written to disk by the file system, > >(because of memory corruption, implementation error, > >the shade of the moon, whatever). > >how could the block device (drbd) know that this data is "invalid"? > >how should it go about "fixing" it, in case it could? > > Checksums all the way down. Solaris' ZFS is pretty cool that way ;-). > But I'm not sure that it would be a good idea for drbd to start doing > things like that (although ZFS on top of drbd would be cool...perhaps > someday). zfs may be cool, sure. but. well. we are a block device. file system feeds us data. we write it. we could checksum _what_ exactly? later file system figures (e.g. by means of checksums) that the data written was invalid, sorry, but dear block device, could you not just give me back the data I meant to write in the first place, pretty please and thank you very much? it does not really work that way. the "traditional note" about backups in the other posting is the exact right answer. -- : Lars Ellenberg Tel +43-1-8178292-0 : : LINBIT Information Technologies GmbH Fax +43-1-8178292-82 : : Schoenbrunner Str. 244, A-1120 Vienna/Europe http://www.linbit.com : __ please use the "List-Reply" function of your email client.