On Fri, Jun 18, 2004 at 05:01:04PM +0200, Michael Jürgens wrote: > > I have a big problem with a RH8 Server. > I installed fedora glibc rpm on a redhat 8.0 by mistake. > And now it is impossible to start rpm updates anymore. > rpm exits with segmentation fault. > > unfortunally I have no access to the console of the server (it is in a > data center far away). I have to make all installations at runlevel 3. > > I changed the following rpms: > glibc-2.2.93-5.i686.rpm to glibc-2.3.2-101.i686.rpm > glibc-common-2.2.93-5.i386.rpm to glibc-common-2.3.2-101.i386.rpm > glibc-devel-2.2.93-5.i386.rpm to glibc-devel-2.3.2-101.i386.rpm > memprof-0.5.0-2.i386.rpm to memprof-0.5.1-3.i386.rpm > binutils-2.13.90.0.2-2.i386.rpm to binutils-2.14.90.0.6-3.i386.rpm > new: tzdata-2003d-1.noarch.rpm > > Is there any chance? Do you have the redhat-8 RPMs correctly installed on any system? At runlevel 3 can you transfer files with scp, ftp, whatever to the troubled box? Push or pull it does not matter. If so, you might try transferring the correct libraries to the system into a handy but non standard place. Then set the environment variable "LD_RUN_PATH" to first search your non standard place for run time libraries. Then run rpm to correct/downgrade the RPMs in question. Once you start updating things with an alternate path for ld do not log out until you are done or have verified that you can make a fresh connection. Put as much in place as possible then begin changes. Even if you only have a FCn box for reference you might use the alternate root option for rpm to unpackage the 'good' RH8 rpms in a handy place so you have good bits to transfer. Also rpm2cpio then cpio. Consider NFS. If you NFS mount a healthy RH8 system and adjust $LD_RUN_PATH and $PATH (see the ld man page) you may also make progress. Caution default NFS flags try and protect root, RTFM. Using the same tactic you might assemble a CDROM that accomplishes the same thing. All these tricks avoid touching the libraries that are keeping the system up at runlevel 3 as long as possible. If you run rpm on a healthy RH8 system with strace you can see each library that it loads. With that information, you might be able to replace file by file, library by library only the stuff that is causing the problem for rpm. And clearly if you have a local spare banger box you can do the same 'wrong' thing then debug and test solutions locally. Good luck, Tom -- T o m M i t c h e l l /dev/null the ultimate in secure storage.
