[CentOS] pup problem

MHR mhullrich at gmail.com
Wed Jun 2 16:19:07 UTC 2010


On Wed, Jun 2, 2010 at 7:02 AM, Charles Campbell
<Charles.E.Campbell at nasa.gov> wrote:
> Hello!
>
> On May 28, I had a problem with pup and seamonkey.  I was doing an
> update via pup and reading a page via seamonkey; seamonkey froze with a
> plain brown screen.  I waited awhile, and then did the
> press-the-power-button-until-it-shuts-off.   However, I don't know
> whether it was seamonkey or pup that actually caused the freeze, but
> apparently pup did not do its updating job correctly.  There's a number
> of icons that are missing, so (for example) open office won't start from
> the menu.  Centos starts, but the desktop background doesn't appear
> (although I can select desktop background and get it to appear).
> Finally, attempts to attach anything to an email via seamonkey (17 or
> 2.04) or thunderbird show the requestor momentarily and then crashes
> seamonkey/thunderbird.
>
> Is there a log somewhere of what was updated (I didn't see it in
> /var/log/messages) and can I force pup to do the update manually?
>
> Any other suggestions for getting things back to normal?
>

For future reference, before cycling the power on a machine with a
hung process, I would suggest:

1) Wait at least fifteen minutes before you try anything.  If the
machine is not completely hogtied, this sometimes allows whatever is
stuck to finish.  Sometimes it's just a matter of patience.

2) Attempt to close the window of the hung process.  Use "Force Quit" as needed.

3) Use the "kill" command if the above does not work.  Use "kill -9"
if a plain kill does not work.

4) If the hung process refuses to go away, run "top" or some other
process monitor to see if any other critical processes are running.
In particular, if you are doing an update, go out of your way not to
interrupt it except as a last ditch desperate resort.

5) Try using one of the consoles to find out which process is hung and
kill it from there (<ctl><alt><F[1-6]> should do this for you;
<ctl><alt><F7> brings you back to the graphics console).

6) If none of that works, try logging out of your gdm.
<ctl><backspace> is an emergency escape method for this if the GUI or
command lines don't work.

7) Only if none of the above gets you anywhere should you reboot.


I don't use pup to update (ever) - I prefer to use yum directly, but I
don't think I've ever seen this particular problem.  Seamonkey has its
own peculiar characteristics, and I've seen it hang many times.
Usually, I just kill it and it comes back just fine.  I have a shell
script for ensuring a complete kill for the hard cases (for both
seamonkey and firefox, but not both at once) where Force Quit doesn't
work:

#!/bin/bash

# A shell script to kill that annoying runaway seamonkey that won't die

case `basename $0` in
        "seakill")      cmd=seamonkey;;
        "foxkill")      cmd=firefox;;
        *)                      echo "Unrecognized command."; exit 1;;
esac

kill -9 `ps -ef | grep $cmd | grep -v grep | tail -1 | awk '{print $2}'`
ps -ef | grep $cmd | grep -v grep

If it works, nothing is displayed.  If seamonkey/firefox is already
gone, it give me kill's error for not finding the process (or for a
missing process number because 'ps' couldn't find it, either).

HTH

mhr



More information about the CentOS mailing list