Ignoring all advice (including my own) I decided that
rather than reconfig and add in I would just upgrade.

When it came time for updates at the end of install,
I picked everything that seemed likely to be on the
system (some things I really had no use for -- how
it got installed I do not know, no package choices
on an upgrade), and got a new kernel, among other
things.

This hosed grub, and the update kernel was not
properly installed, but I untangled a bit of that.

Then I tried to run MCC to pick updates missed and
maybe get the new 22.9.*2mdv kernel a second time
for better luck. Tried to set up sources, and the
thing broke.

When I go to urpmi, I get error messages
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery

but I cannot figure out how to run recovery. recovery on
the command line does nothing. urpmi.recover does not exist
on my system, though mentioned at the bottom of the urpmi
man page.

rpm --rebuilddb yields the same:
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery
error: Runnning db->verify ...
rpmdb: PANIC: fatal region error detected; run recovery
error: db4 error(-30975) from dbenv->open: DB_RUNRECOVERY: Fatal
error, run database recovery

How do I "run database recovery" ?

Cheers!

jim b.
--
UNIX is not user-unfriendly; it merely
expects users to be computer-friendly.

Jim Beard wrote:

There is an item in the Errata that may help (RPM database errors
after upgrading):
http://snipurl.com/1u5ie

I found that by googling
mandriva "run database recovery"
and one of the suggestions pointed to the Errata.

Scott B. wrote:

Yes, maybe I am fumble-fingered or the search at Google
is very sensitive to words/sequence for some reason. I found
old instructions for rm __db* and then rpm -rebuilddb and the
combination worked. briefly.

I went into MCC, and it told me I had 222 packages to update.
After that had been in process for a good while, I walked away
for a few minutes, and came back to find MCC had disappeared.
Restarted it, tried to update, and the problem recurred.

Just removed the __db* files and tried again. No joy.
Removed the __db* files and rpm --rebuilddb again. This
time MCC said I had 56 or 58 update packages. Started that.

MCC blew away again. I repeated rm and rebuild and tried again,
and I had 43 packages to update. Then repeat all, and 24
packages to update. At this point, I tried BitTwisters line
urpmi --wget --auto-update --auto-select --auto
and that got everything but the newest kernel and three packages
uninstallable (lilo was one; I use grub, but at one point had
switched to lilo -- seems no I cannot update the idle lilo).

Used MCC to uninstall vmlinuz-2.6.22.9-desktop-2mdv because
it did not have an initrd-2.6.22.9-desktop-2mdv.img file
and re-installed that. Installed the devel package for
that so nVidia drivers could be compiled. Rebooted.

Miracle of miracles, the nVidia drivers compiled on reboot,
the newest kernel is running, X11 has yet to crash on the
newest kernel (three crashes on the 2.6.22.9-desktop-1mdv
kernel during its brief period of use). Knock on wood
(taps head with knuckles....)

Cheers!

jim b.

--
UNIX is not user-unfriendly; it merely
expects users to be computer-friendly.

On 2007-11-25, Jim Beard wrote:

Quote:

Scott B. wrote:
Jim Beard wrote:

[...]

Quote:

Miracle of miracles, the nVidia drivers compiled on reboot,
the newest kernel is running, X11 has yet to crash on the
newest kernel (three crashes on the 2.6.22.9-desktop-1mdv
kernel during its brief period of use). Knock on wood
(taps head with knuckles....)

Cheers!

jim b.

I was getting random shut-downs a few minutes after logging in to a KDE
session on Mdv2008 - but not with IceWM; I'd assumed the problem lay
somewhere in KDE, and I've noticed a slew of updates for KDE stuff. But I
use XFce4 anyway so I hadn't looked into the matter any further. Next
time I boot my Mdv2008 system I'll try KDE again to see what happens now
that I have the newer kernel.

I don't switch from a working system to a new one all in one go; I
dual-boot and muck around with the new system until I'm happy with it
before making the change-over. I don't have a seperate /home partition
for any system, but I do have a seperate partition for "data" which is
accessible from all the systems I have installed.

--
-- ^^^^^^^^^^
-- Whiskers
-- ~~~~~~~~~~

Whiskers wrote:

I'm not sure whether it was the install kernel or the inability to
get the updates in one fell swoop that created the problems, but
I am suspicious of the kernel. I'm still looking for updates needed
that were not identified/installed in earlier attempts. I remember
seeing a bunch of alsa stuff, and I don't remember seeing all of it
in the things downloaded, some does not appear in /var/log/rpmpkgs,
and updates does not show it as still needed. I will have to go
through packages available for install & select anything that looks
like what I saw listed as updatable and get it.

This is a difference of personality, I think. I do plenty of
tinkering with my toy, perhaps a hold-over from childhood when
Tinker-Toys were a fav of mine, but when it comes to a new
version of the OS I either go for a clean install and have to
tinker with everything to make it work the way I like, or for
an upgrade and have to tinker with things to make it work....
The latter is usually easier, with the drawback that I never
know for sure if my tinkering is complete until things have been
running for weeks without problem.

There are a few particular considerations. I keep two or
three complete backups of everything needed on the system on
a second hard drive, and add a new one before the upgrade/clean
install. If all else fails, I can easily reinstall the old OS
(minimal version) and overwrite everything from backup.

It also helps that the only things I really value on the system
are accumulated data that not only is in backup but on separate
partitions (not reformatted for the new version) and in permanent
storage on CD. Total recovery of everything worth keeping after a
total loss of everything on the system is simply a matter of a
modest amount of time.

Cheers!

jim b.

--
UNIX is not user-unfriendly; it merely
expects users to be computer-friendly.

Jim Beard wrote:

Perhaps I celebrated too soon. This morning, an attempt to
look for files (actually updates) to install ran into the
rpm database problem.

Had to go into /var/lib/rpm and rm __db* and then run
rpm --rebuilddb. That took a few minutes, and then I
could fire up MCC and use it to look for programs.

Odd note: found the BSD games package, so I once again have
adventure on the system. enter get lamp get keys out w e
s forest forest valley s s open .....

Cheers!

jim b.

--
UNIX is not user-unfriendly; it merely
expects users to be computer-friendly.