Update RedHat 6.2 to recent Fedora Core?

Is it practical to run the Fedora Core 4 installer to update a Red Hat
6.2 system? I realize that's a pretty old release, but I'm not looking
forward to manually and meticulously transferring over users, groups,
accounts, etc.

Rick.

On Sat, 28 Jan 2006, in the Usenet newsgroup comp.os.linux.admin, in article
<caedncqIxLcuoEHenZ2dnUVZ_v2dnZ2d@comcast.com>, Rick Ingham wrote:

>Is it practical to run the Fedora Core 4 installer to update a Red Hat
>6.2 system?

I suppose it might work to an extremely limited definition of "work". But
I'm not sure of that many applications that are even similar.

>I realize that's a pretty old release, but I'm not looking forward to
>manually and meticulously transferring over users, groups,
>accounts, etc.

A lot depends on how the system is arranged. Even at home, I keep all of
the user directories on a separate partition. A copy of /etc/passwd,
/etc/group, /etc/shadow (and /etc/gshadow if it exists) will handle the
UNIX side of things (I don't use Samba). Thus, I just point the installer
at the old system (and swap) partitions, and don't allow it to mess with
the partition with /home/ on it. I also have up-to-date backups in case
anything goes horribly wrong. Once the new system is installed, I log
in and verify that the two users (root and "me") work correctly, and have
a usable desktop, then copy the dot files from the "me" directory over to
the old /home directories, because the "new" installation invariably has
different desktop applications, and so on. At work, we try to keep a
"standard" desktop over each update. so the users aren't completely lost.
But then, at work we have the advantage of test systems we play with before
subjecting the ordinary users to the trauma of an update.

RH6.2... came out on 27 March, 2000, support ended 31 March 2003. It was
a great release, and we transitioned most of our users to RH7.2 or 7.3
in 2002. We still have two or three systems on an isolated network running
6.2, with manual updates to such things as the 2.2.26 kernel.

For what it's worth, FC5 is in beta right now, and is "scheduled" for a
March 15th release.

Old guy

Moe Trin wrote:
> On Sat, 28 Jan 2006, in the Usenet newsgroup comp.os.linux.admin, in article
> <caedncqIxLcuoEHenZ2dnUVZ_v2dnZ2d@comcast.com>, Rick Ingham wrote:
>
>
>>Is it practical to run the Fedora Core 4 installer to update a Red Hat
>>6.2 system?
>
>
> I suppose it might work to an extremely limited definition of "work". But
> I'm not sure of that many applications that are even similar.
>
>
>>I realize that's a pretty old release, but I'm not looking forward to
>>manually and meticulously transferring over users, groups,
>>accounts, etc.
>
>
> A lot depends on how the system is arranged. Even at home, I keep all of
> the user directories on a separate partition. A copy of /etc/passwd,
> /etc/group, /etc/shadow (and /etc/gshadow if it exists) will handle the
> UNIX side of things (I don't use Samba). Thus, I just point the installer
> at the old system (and swap) partitions, and don't allow it to mess with
> the partition with /home/ on it. I also have up-to-date backups in case
> anything goes horribly wrong. Once the new system is installed, I log
> in and verify that the two users (root and "me") work correctly, and have
> a usable desktop, then copy the dot files from the "me" directory over to
> the old /home directories, because the "new" installation invariably has
> different desktop applications, and so on. At work, we try to keep a
> "standard" desktop over each update. so the users aren't completely lost.
> But then, at work we have the advantage of test systems we play with before
> subjecting the ordinary users to the trauma of an update.
>
> RH6.2... came out on 27 March, 2000, support ended 31 March 2003. It was
> a great release, and we transitioned most of our users to RH7.2 or 7.3
> in 2002. We still have two or three systems on an isolated network running
> 6.2, with manual updates to such things as the 2.2.26 kernel.
>
> For what it's worth, FC5 is in beta right now, and is "scheduled" for a
> March 15th release.
>
> Old guy
Thanks. I appreciate the info. I think so much is out of date,
possibly including standard file structures/locations and stuff that
everything needs to go. I think I'm better off biting the bullet and
building a new system from scratch and migrating files and users over
afterward.

I noticed this weekend, it's actually version 6.1, so it's even a little
older than what I stated before. We've never updated anything on it.
It's been rock solid since day 1. Last time I checked, a month or two
ago, it had been up solid for well over a year. I don't think we have
another system in 180+ data centers that's been more reliable or up
longer. It's the only Linux system we have. The rest are now all
Solaris and Windows servers. They all get recycled a few times a year
whether they need it or not (monthly for the Windows machines by policy).

Matter of fact, what spawned the inquiry is that I need to update gcc,
and the system has very little space left. So, it's really due for an
update. But everything is still working perfectly and supporting 10
developers on a plain old Dell 450Mhz P3.

Rick.

On Sun, 29 Jan 2006, in the Usenet newsgroup comp.os.linux.admin, in article
<fqydnZjk3NcxAUDeRVn-ow@comcast.com>, Rick Ingham wrote:

>I think so much is out of date, possibly including standard file
>structures/locations and stuff that everything needs to go. I think I'm
>better off biting the bullet and building a new system from scratch and
>migrating files and users over afterward.

Yeah, that's actually what I was trying to say ;-)

>I noticed this weekend, it's actually version 6.1, so it's even a little
>older than what I stated before. We've never updated anything on it.

6.1 came out in October 1999, and support ended in March 2001. I don't
recall us installing it - we jumped from 4.2 to 6.2, and the 5.2 systems
went to 7.2 or 7.3..

>It's been rock solid since day 1. Last time I checked, a month or two
>ago, it had been up solid for well over a year. I don't think we have
>another system in 180+ data centers that's been more reliable or up
>longer.

The problem about long uptimes is that it means you are not updating the
system for problems. I think there was only one worm that was targeting
it ("luckgo" going after holes in nfs-utils), but that would also have
had wu-ftpd-2.5.0 (2.6.0 was later exploited), and I don't even think it
came with SSH.

>It's the only Linux system we have. The rest are now all
>Solaris and Windows servers.

My condolences. Still, I've been supporting SunOS4 in addition to Slowaris.

>Matter of fact, what spawned the inquiry is that I need to update gcc,
>and the system has very little space left.

Depending how far, you probably also need to update glibc2. I doubt any
of the modern gccs will work with glibc2.1.2. For perspective, FC4 is
glibc-2.3.5. There's been a LOT of water under that bridge.

Old guy

Rick Ingham <rdingham@comcast.net> wrote:

[RH 6.1 box:]

> Thanks. I appreciate the info. I think so much is out of date,
> possibly including standard file structures/locations and stuff that
> everything needs to go. I think I'm better off biting the bullet and
> building a new system from scratch and migrating files and users over
> afterward.

Concur. You'll end up copying over stuff from /home, /opt, /usr/local,
/root, and possibly selected places in /usr (e.g., within /usr/lib) and
/var. Be careful about assuming that users' dotfiles in their home
directories will still function correctly: They might or might not.

Keep a copy of the 6.1 box's /etc tree (I'd just do "tar czf
/tmp/etc-2006-02-01.tar.gz /etc"), and refer to its contents -- but
don't try to reuse them directly -- on the new box. When the new box
has all its contents and services working as needed, do a cutover.