ssh connection refused post yum interrupted update

Linux and Unix-Like Red Hat
1

an update of RHEL 5.1 via yum was interrupted after installing the following packages.

Running Transaction
   Updating  : libgcc                       ##################### [  1/951]
   Updating  : setup                        ##################### [   2/951]
  Updating  : filesystem                    ##################### [  3/951]
  Updating  : kernel-headers                ##################### [  4/951]
  Updating  : pciutils-devel                ##################### [  5/951]
  Updating  : busybox                       ##################### [  6/951]
  Updating  :  xkeyboard-config             ##################### [  7/951]
   Updating  : tzdata                       ##################### [  8/951]
   Updating  : glibc-common                 ##################### [   9/951]
  Updating  : glibc                         ##################### [ 10/951]
error: %post(glibc-2.5-49.i686)  scriptlet failed, signal 2

We rolled back the packages to their original versions but cannot connect any new ssh sessions.

We need access to this box remotely, we have 1 open ssh session to the system and can go to the datacenter to access it locally. Any suggestions or help is GREATLY appreciated. Our linux admin is out for medical reasons and we have limited resources.

2

Does yum-complete-transaction work?

3

hi

this is maybe a bug..

i think a lot of idea with this problem..
you try manul update or manul downgrade with --force option and retry update..but this process can be dangereous..

 
rpm -Uvh --force or -Fvh glibc....

or you can try yum option but you must sure from this option about you want it.

 
--skip-broken  # resolve problem causing remove packages

or

rpm --noscripts

## but this can cause a problem and it is neccessary manuel what is postscript's jobs (for ex remove old setup directorys...and some chechk..i m not sure what is process list..maybe you can see glibc source codes..)
or
you can update glibc by manuel install another new version..

Regard Yucel Gemici

4

we had an open ssh session on the system where the original yup update failed and found that ssh was not running

/etc/init.d/ssh restart solved the problem.

We were able to roll back all the packages via rpm to the originals using rpm -q to identify them and then rpm -e to remove the unwanted versions.

Do you see any issues with this method given the yum process that failed? Do I need to check anything prior to rebooting the system in the future?

5

Updating a thousand packages remotely maybe isn't the smartest thing to do.

But it should work.