freebsd stable - Apr 2005 - Newbie Question About System Update

I've been away from *NIX a few years.  I have been playing with FreeBSD 
for a week or so now with mixed results.  I am using release 4.11 
because for some reason 5.3 has problems seeing my hard drives.  4.11, 
Red Hat Linux and NetBSD have no such trouble.

This afternoon I used the "Updating Sources with CVSup" in the FreeBSD
Cheat Sheets and everything worked as advertized.  I believe that it 
advised against using "make world" and suggested that I use
"19.4.1 The
Canonical Way to Update Your System" in the Handbook.  I went through 
the following steps with no problem:

 # make buildworld
 # make installworld
 # mergemaster
 # reboot

After that, I ran into problems.  It took me a little while to figure 
out how to do "boot -s".  However, it appears that a lot of the 
directories aren't mounted and the next scripts aren't in the path.  For
example, I can't figure out how to do the "mergemaster -p".

Any help would be much appreciated.

Jim Campbell

On Monday 18 April 2005 04:51 pm, Jim Campbell wrote:
> I've been away from *NIX a few years.  I have been playing with FreeBSD
> for a week or so now with mixed results.  I am using release 4.11
> because for some reason 5.3 has problems seeing my hard drives.  4.11,
> Red Hat Linux and NetBSD have no such trouble.
>
> This afternoon I used the "Updating Sources with CVSup" in the
FreeBSD
> Cheat Sheets and everything worked as advertized.  I believe that it
> advised against using "make world" and suggested that I use
"19.4.1 The
> Canonical Way to Update Your System" in the Handbook.  I went through
> the following steps with no problem:
>
>  # make buildworld
>  # make installworld
>  # mergemaster
>  # reboot
This is how I do it:
>  # make buildworld
>  # make buildkernel
>  # make installkernel
>  # reboot
>  # make installworld
>  # mergemaster
>  # reboot
-Mike

Many thanks to all of you who very promptly gave me the information 
necessary to progress beyond this point.  Even at 70, I continue to learn.

I never know when I subscribe to a new list how newbie questions will be 
received.  Based on this small sample, I am quite favorably impressed.

Thanks again,

Jim Campbell

> Many thanks to all of you who very promptly gave me the information
> necessary to progress beyond this point.  Even at 70, I continue to learn.
Very impressive! Good luck.

regards
Claus

Jim Campbell <jim-c@charter.net> writes:
>After that, I ran into problems.  It took me a little while to figure 
>out how to do "boot -s".  However, it appears that a lot of the 
>directories aren't mounted and the next scripts aren't in the path. 
For
>example, I can't figure out how to do the "mergemaster -p".
You don't have to do it in single user mode, I never did.  I don't know
why it is recommended that one boots in single user in the Makefile,
perhaps to get a quiescent system without any users and services that
would interfere.  But that can also be achieved by stopping the
high-volume services on the machine after booting, and on a personal
machine (workstation PC) it doesn't matter anyways.  Often it's not even
possible to boot into single-user, for example if you don't have
physical control over the machine (like in a co-lo situation).

mkb.

Jim Campbell <jim-c@charter.net> wrote:
> I've been away from *NIX a few years.  I have been playing with FreeBSD
> for a week or so now with mixed results.  I am using release 4.11 
> because for some reason 5.3 has problems seeing my hard drives.  4.11, 
> Red Hat Linux and NetBSD have no such trouble.
> 
> This afternoon I used the "Updating Sources with CVSup" in the
FreeBSD
> Cheat Sheets and everything worked as advertized.  I believe that it 
> advised against using "make world" and suggested that I use
"19.4.1 The
> Canonical Way to Update Your System" in the Handbook.  I went through 
> the following steps with no problem:
> 
>  # make buildworld
>  # make installworld
>  # mergemaster
>  # reboot
This is not correct, and this is not what 19.4.1 says.  The correct
procedure is as Mike Schultz described.  Please review that section of
the handbook.

If you did, indeed, do as you described, then you have a world that's
out of sync with your kernel.  Try this:
1) Boot in to single user mode
2) fsck
3) mount -a
4) cd /usr/src
5) make buildkernel
6) make installkernel
7) reboot

If you're unable to complete those steps, then you may be better off
reinstalling and trying again - write it off as part of the learning
process.  There are ways to restore your system if you've made this
mistake and the above doesn't work, but it's rather advanced stuff.

-- 
Bill Moran
Potential Technologies
http://www.potentialtech.com

From: Bill Moran <wmoran@potentialtech.com>
Subject: Re: Newbie Question About System Update
Date: Tue, 19 Apr 2005 16:32:37 -0400
> Chuck Swiger <cswiger@mac.com> wrote:
> > Bill Moran wrote:
> > > The system can not replace programs that are in use,
> > 
> > This is generally not the case.  Unix lets you continue to access a
file after
> > it has been deleted, so long as the process hangs on to a file
descriptor.
> > This lets you replace programs in use, without running into the same
problems
> > that platforms like Windows have.
> 
> What you say?:
> 
> bash-2.05b$ su
> Password:
> bolivia# cp /usr/sbin/cron /home/wmoran/.
> bolivia# cp /home/wmoran/cron /usr/sbin/.
> cp: /usr/sbin/./cron: Text file busy
> bolivia# 
mv /usr/sbin/cron /usr/sbin/cron-
cp /blah/cron /usr/sbin/cron

install does this behind the scenes. 

Warner

On Tue, 19 Apr 2005 16:34:37 -0600 (MDT)
Warner Losh <imp@bsdimp.com> wrote:
> From: Bill Moran <wmoran@potentialtech.com>
> Subject: Re: Newbie Question About System Update
> Date: Tue, 19 Apr 2005 16:32:37 -0400
> 
> > Chuck Swiger <cswiger@mac.com> wrote:
> > > Bill Moran wrote:
> > > > The system can not replace programs that are in use,
> > > 
> > > This is generally not the case.  Unix lets you continue to access
a file after
> > > it has been deleted, so long as the process hangs on to a file
descriptor.
> > > This lets you replace programs in use, without running into the
same problems
> > > that platforms like Windows have.
> > 
> > What you say?:
> > 
> > bash-2.05b$ su
> > Password:
> > bolivia# cp /usr/sbin/cron /home/wmoran/.
> > bolivia# cp /home/wmoran/cron /usr/sbin/.
> > cp: /usr/sbin/./cron: Text file busy
> > bolivia# 
> 
> mv /usr/sbin/cron /usr/sbin/cron-
> cp /blah/cron /usr/sbin/cron
> 
> install does this behind the scenes. 
I suppose I have to stand corrected.

Thanks for putting me straight.

-- 
Bill Moran
Potential Technologies
http://www.potentialtech.com

> Apparenlty, nobody who is claiming this has _tried_ it.  Try it yourself
Apparently yo misunderstood what people were claiming. Nobody said
you can modify a running binary, merely that you can replace it. try
deleting it and writing a new one. it does work, I do this all the time!

-pcf.

>> Fact is, trying to update a running system could
result in silent failures.
>> The system can not replace programs that are in
use, so there's always the
>> chance that something or other won't get updated
(cron would be an excellent
>> example ... do you always shut cron off when you
update?  How about
>syslogd?)
>Actually, it can.  install goes to great lengths to
make sure that it
>carefully moves the executable out of the way before
replacing it.  It
>won't go away until the last process to be executing
out of it goes
>away.
>> That being said, I quite often do installworld on
running systems because I
>> have no way to go to single-user mode.  It almost
always works well enough
>> for my purposes, but I don't want anyone to think
that it's "OK" to do this,
>> as it's not guaranteed to work, and will most
likely result in some programs
>> not being updated (such as the examples in the
previous paragraphs).
>It usually works well enough most of the time.  I do
it all the time
>on my development machines.  The problem is "well
enough" and "most of
>the time."
>Warner
I, for one, can attest to that.

To the OP, the thing to watch out for is that just
make sure that the machine isn't busy enough (stop all
daemons under /usr/local/etc/rc.d/, under /etc/rc.d,
just so that the machine's resources are being spent
on the upgrade process), then do the make world dance.

I've done it remotely a lot of times, the thing is you
just have to simulate the process locally(if you have
some spare resources, perhaps a VM?), to avoid
potential problems. Here's a snippet from my one of my
notes back in the 4.X days:

++++++++++++++++++++++

Upgrade path from 4.4Rp47 to 4.10R

       Issues to deal:

       1) SSH Daemon
       Create the user/group
       # pw groupadd -n sshd -g 22
       # pw useradd -n sshd -u 22 -g sshd -c "Secure
Shell Daemon"
-md /var/empty -s /sbin/nologin

       /var/empty must be:
       # ls -al /var/
       dr-xr-xr-x   2 root    wheel   schg  512 Oct 27
 2003 empty

       2) Sendmail
       a) Create the smmsp user/group
       # pw groupadd -n smmsp -g 25
       # pw useradd -n smmsp -u 25 -g smmsp -c
"Sendmail Submission
User" -md /var/spool/clientmqueue -s /sbin/nologin

       make sure that /var/spool/clientmqueue and
everyting under is
owned by smmsp:smmsp and rwx by this user/group
               root@beastman:/var/spool# dir
/var/spool/clientmqueue/
               rw-------   1 smmsp  smmsp  -  97 Jun
22 15:53 sm-client.pid
       If NOT:
       # chown -Rv smmsp:smmsp /var/spool/clientmqueue
       # chmod 770 /var/spool/clientmqueue

       b) Create the mailnull user/group
       # pw groupadd -n mailnull -g 26
       # pw useradd -n mailnull -u 26 -g mailnull -c
"Sendmail
Default User" -md /var/spool/mqueue -s /sbin/nologin
               root@beastman:/var/spool# dir
/var/spool/
               drwxr-xr-x   2 root   daemon  - 512 Jun
23 03:02 mqueue
               drwxr-xr-x   2 root   daemon  - 512 Jun
22 03:01 mqueue.in
               drwx------   2 root   daemon  - 512 Oct
27  2003 opielocks
               drwxr-xr-x   3 root   daemon  - 512 Apr
30 01:06 output

       Aliases file (/etc/mail/aliases):
       Add these:
       mailnull: postmaster
       smmsp: postmaster
       sshd: root

+++++++++++++++++++++++++++

Once the above issues where addressed, the
build/install world went fine.

This was when I was in a situation that I needed to
install a port and wouldn't install on a 4.4R ( name
of port escapes me ;-)



cheers
mars
PS My first email via my gmail account was blocked:
+++++++++++++++++++++++++++++++
This is an automatically generated Delivery Status
Notification

Delivery to the following recipient failed
permanently:

    freebsd-stable@freebsd.org

Technical details of permanent failure:
PERM_FAILURE: SMTP Error (state 10): 554 Service
unavailable; Client host [64.233.162.201] blocked
using bl.spamcop.net; Blocked - see
http://www.spamcop.net/bl.shtml?64.233.162.201



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com