CentOS - Nov 2005 - Beware - Yum 3.5 to 3.6 upgrade replaces named.conf

You get so used to yum upgrades going so smoothly but
I learned the hard way to always make a thorough
inspection after a yum update.  I let yum go ahead and
upgrade from 3.5 to 3.6.  Afterwards I made some basic
queries to httpd, postfix and bind named (probably a
cached query). I even checked the /var/named/
directory and saw all my hosts files.

So looked like another smooth ride, well until today I
noticed my domains dropped off the net. Too bad I did
not check the bind gui or look at /etc/named.conf
after the yum update since it was replaced with a
generic version from 3.6.

Luckily I copied /etc/named.conf-rpmsave to named.conf
and I am back in business, well at least in 24 to 72
hours for the rest of the world :(






		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com

Michael Rock <mikerocks65 at yahoo.com> wrote:
> So looked like another smooth ride, well until today I
> noticed my domains dropped off the net. Too bad I did
> not check the bind gui or look at /etc/named.conf
> after the yum update since it was replaced with a
> generic version from 3.6.
> Luckily I copied /etc/named.conf-rpmsave to named.conf
> and I am back in business, well at least in 24 to 72
> hours for the rest of the world :(
Has nothing to do with YUM.
It has to do with RPM.

If the changes in a package are significant enough, then any
existing configuration files are renamed ".rpmsave" and new
ones take their place.

If the changes are not significant enough that the existing
configuration files will work, the new ones are added with
".rpmnew".

This is how it has been for a long, long time in the RHL
world.

Now I agree you should _never_ see this out of a RHEL/CentOS
update.  That's the whole purpose of backporting changes in
RHEL/CentOS updates -- to _avoid_ this from every occurring.

But apparently it did.


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

On 11/15/05, Michael Rock <mikerocks65 at yahoo.com>
wrote:
> You get so used to yum upgrades going so smoothly but
> I learned the hard way to always make a thorough
> inspection after a yum update.  I let yum go ahead and
> upgrade from 3.5 to 3.6.  Afterwards I made some basic
> queries to httpd, postfix and bind named (probably a
> cached query). I even checked the /var/named/
> directory and saw all my hosts files.
>
> So looked like another smooth ride, well until today I
> noticed my domains dropped off the net. Too bad I did
> not check the bind gui or look at /etc/named.conf
> after the yum update since it was replaced with a
> generic version from 3.6.
>
> Luckily I copied /etc/named.conf-rpmsave to named.conf
> and I am back in business, well at least in 24 to 72
> hours for the rest of the world :(
>
You mean the named.conf file that tells you to use named.custom instead?
This has been filed with RH several times as a bug, and has been
closed repeatedly as NOTABUG. See
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=145244 or
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=145094 for more
details.

Whether you agree with it or not, it appears to be how RH will do
things for the forseeable future.


--
Jim Perrin
System Architect - UIT
Ft Gordon & US Army Signal Center

} Michael Rock wrote:
}
} You get so used to yum upgrades going so smoothly but
} I learned the hard way to always make a thorough
} inspection after a yum update.  I let yum go ahead and
} upgrade from 3.5 to 3.6.  Afterwards I made some basic
} queries to httpd, postfix and bind named (probably a
} cached query). I even checked the /var/named/
} directory and saw all my hosts files.
} 
} So looked like another smooth ride, well until today I
} noticed my domains dropped off the net. Too bad I did
} not check the bind gui or look at /etc/named.conf
} after the yum update since it was replaced with a
} generic version from 3.6.
} 
} Luckily I copied /etc/named.conf-rpmsave to named.conf
} and I am back in business, well at least in 24 to 72
} hours for the rest of the world :(
} 

interesting, on a CentOS 4.1 chroot nameserver i did a

yum upgrade yum

"only" here some time ago and did not experience this issue.

i have not done any upgrades since 4.2 was released other than the yum
upgrade though.

this is just FYI

 - rh

--
Robert Hanson - Abba Communications
   Computer & Internet Services
 (509) 624-7159 - www.abbacomm.net