Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
I'm not don't know all the details all that well, but the short version is basically this: Most filesystems expect to be completely in charge of their block device when they have it mounted. If something else modifies the block device that they have mounted, you can expect complete mayhem. The filesystem will likely go into a faulty disk/corrupt filesystem mode where it goes read only so no more damage can be done to the disk. It might just crash itself, or the entire box outright. OCFS2 (and gfs) are designed to share a block device with other copies of itself running on other systems. Because they're designed with this in mind, they can prevent the mayhem by using locking mechanisms to make sure both machines don't try to mess with the same data at the same time. They can also modify they way they do caching so that they don't make assumptions about what they would have found on the disk, since it may have been changed by another machine. On Thu, 2007-02-22 at 12:46 -0500, David wrote: > Hi, > > I am trying to create a setup where I can read/write to the master node > by samba and read-only, again by samba, from the slave at the same > time. From what I understand I need to use a shared file system and 8.0 > for this. What I'm not clear on is why I need to use this combination. > Is it because other journaled file systems (xfs, reiserfs, etc) cannot > handle drbd updating the underlying data? Also, why is it the > drbd/ocfs2 (for example) combo works? > > If there are already docs explaining all this, please feel free to just > fire off the links to me. > > Thanks. > > > David Filion > > _______________________________________________ > drbd-user mailing list > drbd-user at lists.linbit.com > http://lists.linbit.com/mailman/listinfo/drbd-user