RPM broken
James Olin Oden
joden at lee.k12.nc.us
Sun Apr 4 02:12:23 UTC 2004
On Sat, 3 Apr 2004, Gene C. wrote:
> On Saturday 03 April 2004 12:50, James Olin Oden wrote:
> > On Sat, 3 Apr 2004, Martin Alderson wrote:
> > > I installed fedora core 2 test 2 this morning, and some tool locked up
> > > while using RPM. Now RPM has died - I type any command in and it just
> > > sits there with no output.
> > >
> > > I seem to remembering having to clear a cache or lock file, or maybe
> > > even rebuilding the RPM database.
> > >
> > > Any suggestsions?
> >
> > First off, get the output from the back trace. To do this find rpm's
> > pid, and then run:
> >
> > gdb -p $pid
> > bt # This is in gdb
> >
> > Send that output to the list.
>
> No! Please do not send to the list. File a bugzilla report of your problem
> and add the backtrace there.
>
Let me rephrase this. I am hunting this bug down with Jeff Johnson
(though, he has figured out much more than me) and I want to see the
back trace. If its the traditional rpmdb locks no bug report is required
nor wanted, cause it will probably be a nofix. If it is this new lockup
with scriptlets, then I very much care, and there is already a bugzilla
report that may need to be re-opened. So, if you don't want to see this
sort of stuff on the testers list fine, but it would be very much
acceptable to send it to rpm-list at redhat.com. The bug I am looking at
is bugzilla #117620. What I wanted to know in particular (and I can tell
from the backtrace, is is this the same bug, rather than have him right
a new bug that would have to be duped.
> However, as described below, you may be able to easily fix you problem.
>
If it is the bug I am on the trail of then the answer is not as simple as
that. Basically, if you kill rpm with a -9 and then remove the __db*
files (only because you just had to kill rpm with a -9), then the next
rpm transaction is not likely to lockup. The problem has not gone away,
though, its just that its a tricky little race condition as near as I can
tell such that the problem is still there waiting to rear its head when
you really don't want it to.
Cheers...james
More information about the fedora-test-list
mailing list