[NTLUG:Discuss] Disaster recovery

[Bryan J. Smith]

Leroy Tennison <leroy_tennison at prodigy.net> wrote:
> Along the lines of the meaning of DR, one issue that no one
> raised is a "partial disaster" - the hardware is intact but
> the system won't boot or the service won't start.  In this
> case what would you document?  What came to mind for me was
> menu.lst/grub.conf/lilo.conf and the critical service 
> configuration files.

This is why I _always_ build a boot CD-R/DVD-R for every
server.  I also ensure it is updated monthly as an "end of
month" checklist.

> Another issue I didn't see (or maybe this is the reason for
> the "all of /etc" statements) is special features in use 
> such as quotas or Extended ACLs.  Would a tape restore 
> adequately handle these or is documentation called for?

Well, there is Jorg's "star", which is far more "Austin
Group" (IEEE POSIX 2001+, X/Open Single UNIX Specification,
SUS, v3) compliant than GNU Tar (which is finally just
getting some support in 1.14+).  It can backup some extended
attributes for Ext3 (among other filesystems/OSes).

But this "need" is the #1 reason why I advocate XFS for data
volumes (I still use Ext3 for system volumes).

XFS was designed to _natively_ store such meta-data (upto
64KiB) per inode, and does not use "hidden files" or other
"post-hacks" unlike other filesystems.  It's xfsdump program
then dumps a complete set of inode/meta-data records with
this information, in a way that is compatible with any XFS
implementation on Linux (as well as Irix, which it is
unchanged from in port).

I know everyone hypes up ReiserFS, but it hasn't really cared
about stablizing these features, let alone offered good,
standard POSIX/UNIX interfaces/compatibility (inode, quotas,
etc...).  In fact, most of ReiserFS' compatibility comes from
the general Linux 2.6 virtual filesystem (VFS) layer that
was, not surprisingly, largely SGI code from XFS.

[ SIDE NOTE:  XFS' lack of integration in the stock kernel
2.6 (now backported in late 2.4 releases) was never about
stability, but lack of features in the Linux kernel.  No
joke, the XFS "module" used to be 1MB and that was almost
entirely those "added interfaces" that are now in the 2.6
kernel (although SGI separated out the services, splitting
the module into 2, before that). ]

IBM's JFS was ported from OS/2, instead of AIX for Monterey
Non-Compete reasons (ya know, the _real_ reason why SCO is
suing IBM, and not the post-"damn, IBM didn't settle"
smokescreen of Linux IP).  Hence why JFS for Linux has been
hacked and hacked and hacked in re-adding various, standard
POSIX/UNIX features.  And like ReiserFS, most of JFS' newer
features come from 2.6 VFS features.



-- 
Bryan J. Smith                | Sent from Yahoo Mail
mailto:b.j.smith at ieee.org     |  (please excuse any
http://thebs413.blogspot.com/ |   missing headers)