[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Rebuilding server

From: Eric Siegerman
Subject: Re: Rebuilding server
Date: Wed, 15 Aug 2001 21:43:56 -0400
User-agent: Mutt/1.2.5i

On Wed, Aug 15, 2001 at 06:40:53PM -0600, Ashley M. Kirchner wrote:
>     I'm looking at a server (machine) rebuild.  How can I preserve my
> CVS tree(s)?  I don't know if it's as simple as tarring up my entire
> /CVS directory, then untarring it when I'm done
> (formatting/reinstalling/blah/blah),

Assuming that /CVS is your repository, yes, it is that simple.

Well, almost.  Disable client access to CVS while you back up the
repo, and restore it on the other side of the rebuild.  Those are
rather long critical sections, during which someone might do a
commit, leaving things in an internally inconsistent state.

Personally, I'd do this the bombproof way, by disabling client
access to the entire box.  Bring it down to single-user before
you start taking backups -- of CVS or anything else, including
user directories, mail spools, whatever.

Obviously you'll have to duplicate the CVS configuration on the
(functionally) new server too; don't forget ancillary stuff such
as ssh, (x?)inetd.conf entries, and the like.

(Sorry for the Sysadmin 101 lecture, but better to point this
stuff out than not...)

> I must
> maintain all of the version histories on everything, branches and all.

That's all in the repo; so's the activity log that "cvs history"
reads from.

> I know I can't just checkout everything, then re-import after the
> rebuild because that'll skew the file versioning.

This would be an EXTREMELY bad idea.  It'd lose not just branches
and the details of revision numbers, but ALL old revisions, tags,
commit logs -- everything you're using CVS to preserve in the
first place.  You'd end up with only the most recent revision of
each file, at revision 1.1.


|  | /\
|-_|/  >   Eric Siegerman, Toronto, Ont.        address@hidden
|  |  /
With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea.
        - RFC 1925 (quoting an unnamed source)

reply via email to

[Prev in Thread] Current Thread [Next in Thread]