synchronization/backup mx

[Jack Stone]

Torsten Schlabach wrote:
> > The config you would need is to use drbd to mirror the data
> > directories
>
> Is this officially supported / endorsed?
Not that I know of but someone else may have a better idea than me
> I learned that Cyrus IMAPd (for whatever reason) is quite picky on
> filesystems. From the FAQ:
>
> --- SNIP ---
>
> # Using NFS We don't recommend it. If you want to do it, it may
> possibly work but you may also lose your email or have corrupted
> cyrus.* files. You can look at the mailing list archives for more
> information.
> # Using AFS/Coda We don't recommend it. It's even less likely to work
> than NFS. If you want to do it, it may possibly work but you may also
> lose your email or have corrupted cyrus.* files. CMU's previous e-mail
> system, AMS, leveraged AFS extensively for storage (and transit)
> purposes. For various reasons it didn't scale particularly well and
> led to CMU's interest in IMAP.
>
> Cyrus was designed to use a local filesystem with Unix semantics and a
> working mmap()/write() combination. AFS doesn't provide these
> semantics so won't work correctly.
>
> --- SNIP ----
drbd isn't a file system it works on the disk level so you can have
whatever file system on top of it. You end up with the mirrored disk
partitions on each server being identical. AFAIK it is fully posix
compliant and shouldn't have any of the problems mentioned above.
> The concept of mirroring the BerkeleyDB database files on a filesystem
> level does not sound that reliable to me. I might be wrong ...
I can see your concerns and as haven't tried it I can't comment.
However, it is at a block level with only one node able to write to the
disk so there should be no difference as far as Berkly is concerned.

Regards,

 Jack Stone