Hello,

Yesterday I decided to upgrade to slack11. I was using 10.1 for one year,
upgraded to 10.2 (no problems but gvim was not working) and as I saw that
I made it to 10.2 I tried to upgrade to 11.

But I made (I suppose) a huge mistake.

Full story:
I copied all the packages to one directory from cd slack 11 and after
studying UPGRADE.TXT and CHANGES_AND_HINTS.TXT

I 'v done this:
(as root)
telinit 1
upgradepkg glibc-solibs*.tgz
upgradepkg pkgtools*.tgz
upgradepkg glibc*.tgz
upgradepkg devs*.tgz (in CHANGES_AND_HINTS.TXT some caution about this package)

then with 300Mbytes free space left (my huge mistake?), I upgraded (?!) the
rest of packages. I was confident that my space is enough, because many
packages was skipped (allready installed) so I decided to sleep:
alt+f2 (another terminal)
echo "telinit 3" | at 08:00

error: no atd server running (oops)

atd -b 15 -l 1 (in the meantime upgrade was happening, atd was allready
upgraded)
echo "telinit 3" | at 08:00
(sleep now)

But when I checked the "upgrade" I was terrified about the fact that /
has 0 bytes free!

I started to erase useless stuff like locales so now I have 136Mbytes free.

Modified / erased the .new files from /etc to suit my configuration.

Problem:
iceWM does not work and my monitor is freezed. I logged-in with ssh and I am
writing this.

Also I was thinking that maybe /var/log/messages was fulling my disc. I saw
errors every 2 seconds about gpm (!). I killed it.

But programs DO work and it seems that upgrade was succesfull, for instance...
the packages where upgraded alphabeticaly. So gvim now works (remote X11) and
it was installed on the far end! Lilo is updated, e.t.c.

With NX I can log-on to pwm or KDE everything seems fine, but my physical
monitor is freezed.

My Questions:
1. How can I tell if the upgraded was indeed succesfull?
2. Should I reinstall "by force" the packages?
3. *How can I restore (unfreeze) my monitor?*
4. Any other ideas?

In a identical test machine I have vanilla Slackware 11, (not upgraded, just
installed) no problems there.

I have a backup of /etc /boot /bin /sbin /var /usr (not /opt/kde). Personal
files are on different partition and backed up.
--
Please excuse my english writing!
Slackware 11.console
Knowledge report: Newbie with custom kernel

On Sat, 27 Jan 2007, korgman wrote:

k> then with 300Mbytes free space left (my huge mistake?), I upgraded (?!) the
k> rest of packages. I was confident that my space is enough, because many
k> packages was skipped (allready installed) so I decided to sleep:


On a new install on my new computer, /opt is 688M. Rather than
repartition, I made /opt a symbolic link to /usr/opt as I have /usr on a
different partion.

Of the 688M, 684M is /opt/kde.


--
Alan

On 2007-01-27, Alan Clifford <sardines@purse-seine.net> wrote:
> On a new install on my new computer, /opt is 688M. Rather than
> repartition, I made /opt a symbolic link to /usr/opt as I have /usr on a
> different partion.

I have done this some years ago and KDE has some strange problems. But I moved
OpenOffice2 so I have some extra space.

Problem solved. the problem was xorg.conf! Very lucky that I have identical
machine. I copied the file and everything is ... almost fine. (some minor
problems, I hope I can handle myself)

Thanks for your time Alan.

Just to make sure I will reinstall all the Slack11 packages.

--
Please excuse my english writing!
Slackware 11
Knowledge report: Newbie with custom kernel

On Jan 27, 3:49 am, korgman <kor...@gmail.com> wrote:
> But I made (I suppose) a huge mistake.

Been there. :-) I used to sit there biting my nails and watching df
as the space would slowly shrink away, then grow when old packages
were removed, then shrink again to new and alarming lows.

When I went from 10.1 to 10.2, it was just too much. I repartitioned
the disk.

> 1. How can I tell if the upgraded was indeed succesfull?

When in doubt, I would just force a reinstall.

> 4. Any other ideas?

It wouldn't surprise me if there was some script out there that would
audit all your packages. But I don't know of one off the top of my
head.

For my personal machines (and this is just my holywar opinion, mind
you) I now like to have a single partition for the whole system.
(Except for my home directory which is on its own disk.) No more of
this pesky running out of space on /.

With throwaway disks the huge sizes they are now, I end up with a lot
of free space on /, but I can use that to back up my laptop, or
whatever.

-Beej

On 2007-01-27, Alan Clifford <sardines@purse-seine.net> wrote:
> On Sat, 27 Jan 2007, korgman wrote:
>
> k> then with 300Mbytes free space left (my huge mistake?), I upgraded (?!) the
> k> rest of packages. I was confident that my space is enough, because many
> k> packages was skipped (allready installed) so I decided to sleep:
>
>
> On a new install on my new computer, /opt is 688M. Rather than
> repartition, I made /opt a symbolic link to /usr/opt as I have /usr on a
> different partion.


Same idea here, but this seemed like a cleaner implementation:
In /etc/fstab:
/usr/opt /opt none bind

A side benefit is that "ls /opt" will actually return a listing of the
contents rather than "/opt@" (the symlink itself).

RW

On 2007-01-28, Robby Workman <newsgroups@rlworkman.net> wrote:
> On 2007-01-27, Alan Clifford <sardines@purse-seine.net> wrote:
>> On a new install on my new computer, /opt is 688M. Rather than
>> repartition, I made /opt a symbolic link to /usr/opt as I have /usr on a
>> different partion.
> /usr/opt /opt none bind
Oh! That's better. Thank you, now maybe also move kde to another partition.

After the reinstallation another 100Mbytes are gone. So, I suppose that I was
missing 100MBytes (!) of files.

--
Please excuse my english writing!
Slackware 11
Knowledge report: Newbie with custom kernel

>
> Same idea here, but this seemed like a cleaner implementation:
> In /etc/fstab:
> /usr/opt /opt none bind
>
> A side benefit is that "ls /opt" will actually return a listing of the
> contents rather than "/opt@" (the symlink itself).

yep, this is what I use too (/ was going low on space so I moved /opt to
the /home partition).

The additional benefit is that some applications resolve the symbolic links,
and will find that kde is in /usr/opt/kde and will create config files with
this path. On the other hand a bind mount is undetectable.


--
damjan