RPM database

[Pedro Fernandes Macedo]

I have a little situation here... My raid failed and I lost the /var
directory , among others.. I managed to make the system start again ,
but I need to update it (the / and /boot partitions were raid 1 , but
they werent in sync , dont ask me why..) , because the latest kernel I
have is 2149... but I can update , because the rpm db is gone.. I cant
rebuild the rpmdb , because it tries to use /var/lib/rpm/Packages ,
which is gone also... 
Any ideas on how to rebuild the db?
(I cant reinstall the system now.. lots of things depend on this machine
and I need to do some things before reinstalling it)


--
Pedro Macedo

[Bevan C. Bennett]

Pedro Fernandes Macedo wrote:
>  I have a little situation here... My raid failed and I lost the /var
>  directory , among others.. I managed to make the system start again ,
>  but I need to update it (the / and /boot partitions were raid 1 , but
>  they werent in sync , dont ask me why..) , because the latest kernel I
>  have is 2149... but I can update , because the rpm db is gone.. I cant
>  rebuild the rpmdb , because it tries to use /var/lib/rpm/Packages ,
>  which is gone also... 
>  Any ideas on how to rebuild the db?
>  (I cant reinstall the system now.. lots of things depend on this machine
>  and I need to do some things before reinstalling it) 
/var/lib/rpm/Packages is installed as part of the 'rpm' rpm.
If you grab the .src.rpm you should be able to extract the file and 
re-install it manually. It also -might- work to run "rpm -i --force 
rpm-*.rpm".

If the db is completely gone, you'll probably need to start by creating 
one: "rpm --initdb". At that point you might be able to re-install your 
rpms   with "--justdb" (if all their pieces are still in place) or 
"--force" (if you want the files physically reinstalled).

I've never had to try to rebuild /var from scratch, so I can't really 
say what will and won't work.

[Jason Dixon]

On Mar 4, 2004, at 2:16 PM, Pedro Fernandes Macedo wrote:

>  I have a little situation here... My raid failed and I lost the /var
>  directory , among others.. I managed to make the system start again ,
>  but I need to update it (the / and /boot partitions were raid 1 , but
>  they werent in sync , dont ask me why..) , because the latest kernel I
>  have is 2149... but I can update , because the rpm db is gone.. I cant
>  rebuild the rpmdb , because it tries to use /var/lib/rpm/Packages ,
>  which is gone also...
>  Any ideas on how to rebuild the db?
>  (I cant reinstall the system now.. lots of things depend on this 
>  machine
>  and I need to do some things before reinstalling it) 
Just restore from your backups, then rpm --rebuilddb.

--
Jason Dixon, RHCE
DixonGroup Consulting
http://www.dixongroup.net

[Pedro Fernandes Macedo]

Jason Dixon wrote:

>  Just restore from your backups, then rpm --rebuilddb.
>  
Sadly this was our backup server , and we never made a backup of it ... 
just from the other machines... I found out that I've lost the index 
files for the amanda backups I had , so I can simply reinstall my server 
now...
The odd thing: it was installed on a set of raid 1 partitions... /boot 
and / were different on each disk ( /boot had the old 2149 kernel on one 
disk and the other had only 2166 and 2174 kernels..) I wish I knew why 
this difference , as raid 1 is supposed to be mirroring...

But thanks for the help guys..

--
Pedro Macedo

[Bevan C. Bennett]

Pedro Fernandes Macedo wrote:
>  Jason Dixon wrote:
>  
> >
> > Just restore from your backups, then rpm --rebuilddb.
> >
>  Sadly this was our backup server , and we never made a backup of it ... 
>  just from the other machines... I found out that I've lost the index 
>  files for the amanda backups I had , so I can simply reinstall my server 
>  now...
>  The odd thing: it was installed on a set of raid 1 partitions... /boot 
>  and / were different on each disk ( /boot had the old 2149 kernel on one 
>  disk and the other had only 2166 and 2174 kernels..) I wish I knew why 
>  this difference , as raid 1 is supposed to be mirroring... 
It sounds like the mirror got broken at some point, perhaps when a disk 
error was detected (yielding seven kernel versions of bad luck).

If you don't watch the logs closely, it can be disturbingly easy to miss 
RAID-related messages for things like 'I failed this drive today'.

[Pedro Fernandes Macedo]

Bevan C. Bennett wrote:

>  It sounds like the mirror got broken at some point, perhaps when a 
>  disk error was detected (yielding seven kernel versions of bad luck).
>
>  If you don't watch the logs closely, it can be disturbingly easy to 
>  miss RAID-related messages for things like 'I failed this drive today'. 
Looks like my co-worker didnt see this e-mail then... He receives all 
the e-mails to root for our domain (about 100 linux boxes , and about 50 
sun) and he usually doesnt have the time to read them all.. I only saw 
the failed raid when I changed the mail config on our mailqueue/backup 
server to ensure that our migration to postfix was working perfectly...  
I guess we need to change the default policy for root e-mails...

--
Pedro Macedo