This release fixes a problem where data copied from windows to lessfs (samba) would show the wrong nr of blocks. This would result in du reporting wrong disk usage numbers.
-
Archives
- January 2015
- August 2014
- February 2014
- December 2013
- May 2013
- March 2013
- January 2013
- December 2012
- July 2012
- June 2012
- April 2012
- January 2012
- October 2011
- September 2011
- August 2011
- June 2011
- May 2011
- April 2011
- March 2011
- February 2011
- January 2011
- December 2010
- October 2010
- September 2010
- August 2010
- July 2010
- June 2010
- April 2010
- March 2010
- February 2010
- January 2010
- December 2009
- November 2009
- October 2009
- September 2009
- August 2009
- June 2009
- March 2009
-
Meta
Thank you for the fix! I was getting ready to set up samba shares for backups until I read about the bug. Let me tell you, connecting to NFS shares through windows is no walk in the park! I’m excited to try this out with my samba shares, thank you!
I did an upgrade from 1.0.6 to 1.0.7 and it rolled back my entire filesystem. Is this expected?
Mar 29 00:24:33 sheevaplug lessfs[20984]: The selected data store is file_io.
Mar 29 00:24:33 sheevaplug lessfs[20984]: Lessfs transaction support is enabled.
Mar 29 00:24:33 sheevaplug lessfs[20984]: Lessfs uses a 24 bytes long hash.
Mar 29 00:24:33 sheevaplug lessfs[20984]: Lessfs fsync does not sync the databases to the disk when fsync is called on an inode
Mar 29 00:24:33 sheevaplug lessfs[20984]: cache 2048 data blocks
Mar 29 00:24:33 sheevaplug lessfs[20984]: The tiger hash has been selected.
Mar 29 00:24:33 sheevaplug lessfs[20984]: Rollback : truncate /media/dedupe/dta/blockdata.dta from 121144737140 to 0
Mar 29 00:29:14 sheevaplug mklessfs[21010]: The selected data store is file_io.
Mar 29 00:29:14 sheevaplug mklessfs[21010]: Lessfs transaction support is enabled.
Mar 29 00:29:14 sheevaplug mklessfs[21010]: Lessfs uses a 24 bytes long hash.
Mar 29 00:29:14 sheevaplug mklessfs[21010]: Lessfs fsync does not sync the databases to the disk when fsync is called on an inode
Mar 29 00:29:14 sheevaplug mklessfs[21010]: cache 2048 data blocks
Mar 29 00:29:14 sheevaplug mklessfs[21010]: The tiger hash has been selected.
Hi Brian,
This is very unexpected and of course a rather critical event.
Did you cleanly umount the filesystem before upgrading it?
Have you enabled transactions after you upgraded?
I tested some of the obvious scenario’s but I am unable to replicate this.
Can you tell me exactly what you have done and share your config?
Mark
I have found the problem.
A metadata field is not present in older versions of lessfs. I could reproduce this bug by migrating from lessfs-1.0.0 without transaction support to lessfs-1.0.8 with transactions enabled.
lessfs-1.0.8 fixes this problem by automatically adding the missing metadata when it detects the old structure.
Mark.
It works! The Samba bug is gone.