Re: LMDB integrity with MDB_NOSYNC on various filesystems

[Howard Chu <hyc@symas.com>]

Dimitrios Apostolou wrote:
> Hello,
>
> I'm reading in LMDB docs that MDB_NOSYNC will not allow database
> corruption under certain conditions:
>
> > However, if the filesystem preserves write order and the MDB_WRITEMAP
> > flag is not used, transactions exhibit ACI (atomicity, consistency,
> > isolation) properties and only lose D (durability). I.e. database
> > integrity is maintained, but a system crash may undo the final
> > transactions
>
>
> Do we know if these conditions are valid for modern filesystems like ext3,
> ext4, XFS and how this is affected by various mount options (e.g. ordered,
> writeback etc)?

Read up on the specific filesystems you're interested in. These things change 
too frequently for us to track, and it's your responsibility to know what 
you're deploying.

--
  -- Howard Chu
  CTO, Symas Corp.           http://www.symas.com
  Director, Highland Sun     http://highlandsun.com/hyc/
  Chief Architect, OpenLDAP  http://www.openldap.org/project/