On Wed, Sep 7, 2011 at 6:33 PM, Justin Gottula <jus...@jgottula.com> wrote: > > Hi, > > I recently created a Btrfs volume on top of a software (mdadm) raid5 array > (since Btrfs currently lacks raid5 support at the FS level). On this 640 GB > volume, I stored a ~400 GB tar file. After a couple weeks of use, I used > 'btrfs > defragment' on this file in an effort to (a) defrag and (b) compress the > file. I > made sure I was using the latest version of the userspace utilities (Btrfs > v0.19-35-g1b444cd-dirty) as well as kernel 3.0.
did you use the integration branch, ie: http://git.darksatanic.net/repo/btrfs-progs-unstable.git ... this has the latest code for the time being. looks like `integration-20110805` is the most recent head. > <snip> > > At this point, I took a disk image and dived in, and in doing so discovered > that > somehow, there were CHUNK_ITEMs in the chunk tree that referred to physical > address ranges that were entirely outside of the device (matching up to the > ranges showing up in the kernel log over and over). Evidently, the filesystem > driver thought it should move the root tree onto a chunk that existed at a > nonexistant offset in the device. I checked the superblocks and verified that > the total_bytes fields matched up correctly to the actual device size, which > leaves me wondering how those chunks ever got there. i could be way of base here, but your report reminded me of: [thread] http://www.spinics.net/lists/linux-btrfs/index.html#12121 ----------------------------------------------- extent data backref root 5 objectid 258 offset 18446744073709543424 count 1 extent data backref root 5 objectid 257 offset 0 count 1 > So I think we have to live with this defect, just fix relocation for > the negative offset case ? I prefer fixing relocation. ----------------------------------------------- ... which, if i understood correctly, surfaced some issues with relocation that could cause the offset to be grossly inaccurate (eg. off the device completely?) could of course be completely unrelated :-) -- C Anthony -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html