ARRRRGH! Guys, who's breaking -STABLE's GMIRROR code?!

[Oliver Fromme]

Marc G. Fournier wrote:
 > Steve O'Hara-Smith wrote:
 > > Greg Barniskis <nalists at scls.lib.wi.us> wrote:
 > > 
 > > > If you /track/ STABLE by frequently cvsupping it and rebuilding your
 > > > system, you will very likely encounter a serious problem sooner or
 > > > later. That's why tracking it is not recommended for production
 > > > systems.
 > > 
 > > I did exactly that all the way from 2.0 to 4.11 on various machines
 > > without ever having any trouble.
 > 
 > Ditto ... in fact, I do that on my desktop and have yet to hit a problem 
 > ... -STABLE *is* generally very stable ...

Same here.

However, if you want (or need) to track stable, there are
certain possibilities to avoid trouble.  Of course watching
the -stable mailing list (and possibly even -cvs-all) and
reading /usr/src/UPDATING should be a "must".  But there
are more things that can be done.

On important production machines, it might be a good idea
to track -stable with some delay.  For example, always update
to the -stable date of 4 weeks ago (using the -D option of
cvs, or the "date=" keyword of cvsup), after making sure that
no critical problems have been reported in the mailing list
in the past 4 weeks.  Chances are that critical bugs are
detected and fixed pretty quickly in the -stable branch.

And of course:  Always make sure that you have good backups.
But that's even true if you don't track -stable.

Best regards
   Oliver

PS:  Some people think that a RAID1 (mirror) is a substitute
for a backup.  It's not.

-- 
Oliver Fromme,  secnetix GmbH & Co. KG, Marktplatz 29, 85567 Grafing
Dienstleistungen mit Schwerpunkt FreeBSD: http://www.secnetix.de/bsd
Any opinions expressed in this message may be personal to the author
and may not necessarily reflect the opinions of secnetix in any way.

"A language that doesn't have everything is actually easier
to program in than some that do."
        -- Dennis M. Ritchie