Re: restarting with a new pool

On 31 Aug 2004, Paul Fox wrote:
> is there a quick recipe for re-starting from a new pool?  somehow
> my backuppc filesystem got corrupted -- so i'd like to, essentially,
> do the following:

My suggestion would be to copy the few configuration files to somewhere
safe, remove the *whole* thing, install afresh, then copy the
configuration files back.

That would be easier and less error-prone than trying to prune out
things that you don't want.

[...]

> (secondary question: the previous filesystem for the pool was ext3.
> i'm running a 2.4.24 kernel.  is ext3 the right choice?  i'm not
> sure where my corruption came from -- i had corrupt directories, links
> from files to directories, other bad stuff.  :-/ )

I would strongly recommend that you stick with ext3.  The other three
options are ext2, xfs and reiserfs, pretty much, which have their
drawbacks compared to ext3.

ext2, being non-journaled, rapidly becomes a three hour ordeal while
waiting for a full fsck on an unclean reboot -- not my idea of fun with
the huge size of many backuppc installs.


xfs is metadata journaled only, so your actual data is not protected on
a reboot. If the data itself has not been written on reboot, for
security reasons it is nulled out where the metadata changed.

While they have reduced the windows where this can regularly happen, it
still does, and is the bane of my existence with some XFS-based systems
I have to maintain.

>From experience, expect semi-regular data corruption if you have an
unclean restart of your xfs using system.


reiserfs is probably the best of the alternative filesystems, but
suffers from two major issues:  it still gets semi-regular reports of
corruption problems and stability issues, and the developers have moved
on to the new new thing, version 4.

Personally, I don't use it and the main reason is that I don't trust the
development style that was used. Treating the user-space verification
and testing tools as the lowest, rather than highest, priority, struck
me as the most obvious sign of a development process driven more by
feature achievements than stability.

Others, though, have had good things to say about it, so obviously YMMV.
Also, I have not made recent use of it and, these days, the developers
claim it to be stable and reliable.

      Daniel
-- 
Consumption is a treatable disease.
        -- Tibor Kalman



-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click
_______________________________________________
BackupPC-users mailing list
BackupPC-users@xxxxxxxxxxxxxxxxxxxxx
https://lists.sourceforge.net/lists/listinfo/backuppc-users
http://backuppc.sourceforge.net/